Method for making predictions related to polycondensation reaction, information processing device, and program

Abstract

The purpose is to improve prediction technology related to polycondensation reactions.　Provided is a method for making predictions related to a polycondensation reaction, the method being performed by an information processing device 10, wherein the method includes a step in which a prediction model is trained based on actual data including a plurality of explanatory factors and objective factors related to a polycondensation reaction and a step that predicts by the prediction model objective factors in the polycondensation reaction based on a plurality of explanatory factors related to the polycondensation reaction; the plurality of explanatory factors include a plurality of feature values obtained by clustering analysis of time-series data of a plurality of instruments in a dehydration and heating step; and the objective factors include at least one of the viscosity and acid value.

Description

Method for predicting polycondensation reactions, information processing device, and program

[0001] The present disclosure relates to a method for predicting polycondensation reactions, an information processing device, and a program. This application claims priority based on Japanese Patent Application No. 2023-027827 filed on February 24, 2023, the contents of which are incorporated herein by reference.

[0002] Conventionally, methods for making predictions regarding chemical reactions have been proposed (for example, Patent Document 1).

[0003] WO 2003 / 026791

[0004] The technology described in Patent Document 1 discloses that modeling techniques such as neural networks, partial least squares, and principal component regression are used to optimize the control of a reactor system. However, no consideration is given to specific design methods and optimizations when making predictions related to polycondensation reactions, and there is room for improvement in prediction techniques related to polycondensation reactions.

[0005] In view of the above circumstances, an object of the present disclosure is to improve prediction techniques relating to polycondensation reactions.

[0006] (1) A method according to one embodiment of the present disclosure is a method for making predictions related to a polycondensation reaction, executed by an information processing device, comprising: a step of training a prediction model based on performance data including a plurality of explanatory factors and an objective factor related to the polycondensation reaction; and a step of predicting the objective factor during the polycondensation reaction using the prediction model based on the plurality of explanatory factors related to the polycondensation reaction, wherein the plurality of explanatory factors include a plurality of feature quantities obtained by clustering analysis of time-series data from a plurality of measuring instruments in a dehydration heating process, and the objective factor includes at least one of viscosity and acid value.

[0007] (2) In one embodiment of the present disclosure, the method is the method according to (1), wherein the plurality of explanatory factors include theoretical values ​​in a reaction physics model.

[0008] (3) In one embodiment of the present disclosure, the method is the method according to (1) or (2), in which the polycondensation reaction is a dehydration condensation reaction of a polyester.

[0009] (4) A method according to one embodiment of the present disclosure is a method according to any one of (1) to (3), wherein in the predicting step, a change in the objective factor over time during the polycondensation reaction is predicted in advance, the plurality of explanatory factors include a calculated cumulative value of raw material addition and a raw material addition time during the polycondensation reaction, and the method further varies the calculated cumulative value of raw material addition and the raw material addition time, which are part of the calculation conditions for the predicted value of the change in the objective factor over time, and outputs a visualized graph showing the relationship between the reaction end time, the amount of raw material added, and the raw material addition time.

[0010] (5) In one embodiment of the present disclosure, the method is the method described in (4), wherein the visualization graph is either a heat map or a contour map, with the first axis representing the ingredient addition time and the second axis representing the ingredient addition amount.

[0011] (6) In one embodiment of the present disclosure, the method is the method according to (4) or (5), wherein the visualized graph includes a plot showing performance data.

[0012] (7) In one embodiment of the present disclosure, the method is a method according to any one of (1) to (6), wherein the prediction model is a neural network model including an input layer, an intermediate layer, and an output layer, and the coefficient of the activation function associated with the intermediate layer is greater than the coefficient of the activation function associated with the output layer.

[0013] (8) An information processing device in one embodiment of the present disclosure is an information processing device that includes a control unit and makes predictions related to a polycondensation reaction, wherein the control unit trains a prediction model based on performance data including a plurality of explanatory factors and an objective factor related to the polycondensation reaction, and predicts the objective factor during the polycondensation reaction using the prediction model based on the plurality of explanatory factors related to the polycondensation reaction, wherein the plurality of explanatory factors include a plurality of feature quantities obtained by clustering analysis of time-series data from a plurality of measuring instruments in a dehydration heating process, and the objective factor includes at least one of viscosity and acid value.

[0014] (9) A non-transitory computer-readable recording medium according to an embodiment of the present disclosure is a non-transitory computer-readable recording medium storing instructions, which, when executed by a processor, cause the processor to execute the following: training a prediction model based on performance data including a plurality of explanatory factors and an objective factor related to the polycondensation reaction; and predicting the objective factor during the polycondensation reaction using the prediction model based on the plurality of explanatory factors related to the polycondensation reaction, wherein the plurality of explanatory factors include a plurality of feature quantities obtained by clustering analysis of time-series data of a plurality of measuring instruments in a dehydration heating process, and the objective factor includes at least one of viscosity and acid value.

[0015] According to the method, information processing device, and program for making predictions regarding polycondensation reactions according to an embodiment of the present disclosure, it is possible to improve prediction techniques regarding polycondensation reactions.

[0016] 1 is a block diagram showing a schematic configuration of an information processing device according to the present embodiment; FIG. 2 is a flowchart showing the operation of the information processing device according to the present embodiment; FIG. 3 is a concept showing the process of polycondensation reaction according to the present embodiment; FIG. 4 is a diagram showing the time progression of each piece of data related to the dehydration and temperature rising process; FIG. 5 is a diagram showing the time progression of categories related to the dehydration and temperature rising process; FIG. 6 is an example of a result of accuracy verification of a prediction model according to the present embodiment; FIG. 7 is an example of a visualized graph according to the present embodiment; FIG. 8 is an example of a conceptual diagram of a prediction model in the present embodiment.

[0017] A method, an information processing device, and a program for making predictions related to polycondensation reactions according to embodiments of the present disclosure will be described below with reference to the drawings. Prediction targets of embodiments of the present disclosure include both batch reactions and continuous reactions. Main polymer materials synthesized by polycondensation reactions according to this embodiment include polyester, polyamide, polyethylene terephthalate, urea resin, phenolic resin, silicone resin, alkyd resin, alkyd resin polyether, polyglucoside, melamine resin, polycarbonate, etc. For example, the polycondensation reaction according to this embodiment includes a dehydration condensation reaction of polyester.

[0018] 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.

[0019] First, an overview of this embodiment will be described. The method for making predictions related to a polycondensation reaction in this embodiment is executed by an information processing device 10. The information processing device 10 trains a prediction model based on performance data including a plurality of explanatory factors and an objective factor related to the polycondensation reaction. Furthermore, the information processing device 10 predicts the objective factor during the polycondensation reaction based on the plurality of explanatory factors related to the polycondensation reaction using the trained prediction model. Here, the plurality of explanatory factors include a plurality of feature quantities obtained by clustering analysis of time-series data from a plurality of measuring instruments in the dehydration temperature-raising process. Furthermore, the objective factor is characterized by including at least one of viscosity and acid value.

[0020] As described above, according to this embodiment, the multiple explanatory factors include multiple feature quantities obtained by clustering analysis of time-series data from multiple measuring instruments in the dehydration and heating step. The objective factor is characterized by including either viscosity or acid value. When predicting such an objective factor in a polycondensation reaction, prediction accuracy can be improved by including multiple feature quantities obtained by clustering analysis of time-series data from multiple measuring instruments in the dehydration and heating step, as described below, in the explanatory factors. Therefore, this embodiment can improve prediction technology related to polycondensation reactions.

[0021] (Configuration of Information Processing Device) Next, each component of the information processing device 10 will be described in detail with reference to Fig. 1. The information processing device 10 is any device used by a user. For example, a personal computer, a server computer, a general-purpose electronic device, or a dedicated electronic device can be employed as the information processing device 10.

[0022] As shown in FIG. 1, the information processing device 10 includes a control unit 11, a storage unit 12, an input unit 13, and an output unit 14.

[0023] 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 controls each unit of the information processing device 10 and executes processes related to the operation of the information processing device 10.

[0024] 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.

[0025] The input unit 13 includes at least one input interface. The input interface may be, for example, a physical key, a capacitive key, a pointing device, or a touch screen integrated with a display. The input interface may also be, for example, a microphone that accepts voice input, or a camera that accepts gesture input. The input unit 13 accepts an operation to input data used in 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).

[0026] The output unit 14 includes at least one output interface. The output interface is, for example, a display that outputs information as a video. The display is, for example, a liquid crystal display (LCD) or an organic electroluminescence (EL) 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, such as USB, HDMI (registered trademark), or Bluetooth (registered trademark), can be used.

[0027] 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.

[0028] 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.

[0029] 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.

[0030] In this embodiment, the storage unit 12 stores, for example, performance data and a prediction model. The performance data and the prediction model may be stored in an external device separate from the information processing device 10. In this case, the information processing device 10 may be provided with an external communication interface. The communication interface may be either a wired or wireless communication interface. In the case of wired communication, the communication interface may be, for example, a LAN interface or a USB. In the case of wireless communication, the communication interface may be, for example, an interface compatible with a mobile communication standard such as LTE, 4G, or 5G, or an interface compatible with short-range wireless communication such as Bluetooth (registered trademark). The communication interface is capable of receiving data used in the operation of the information processing device 10 and transmitting data obtained by the operation of the information processing device 10.

[0031] (Operation of Information Processing Apparatus) Next, the operation of the information processing apparatus 10 according to this embodiment will be described with reference to FIG.

[0032] Step S101: The control unit 11 of the information processing device 10 trains a prediction model based on performance data related to the polycondensation reaction. The performance data includes explanatory factors and objective factors related to the polycondensation reaction. The explanatory factors include multiple feature quantities obtained by clustering analysis of time-series data from multiple measuring instruments in the dehydration and heating process. The objective factors include at least one of viscosity and acid value. In other words, the control unit 11 trains the prediction model using these explanatory factors and objective factors included in the performance data as learning data.

[0033] Any method can be used to acquire the performance data. For example, the control unit 11 acquires the performance data from the storage unit 12. The control unit 11 may also acquire the performance data by receiving input of the performance data from a user via the input unit 13. Alternatively, the control unit 11 may acquire the performance data from an external device that stores the performance data via a communication interface.

[0034] The prediction model trained based on the learning data is subjected to cross-validation. If the result of the cross-validation shows that the accuracy is within a practical range, the prediction model is used to make predictions regarding polycondensation reactions.

[0035] Step S102: The control unit 11 predicts the objective factor related to the polycondensation reaction based on the plurality of explanatory factors related to the polycondensation reaction. For example, the control unit 11 may acquire the explanatory factors by receiving input of the explanatory factors from a user via the input unit 13.

[0036] Step S103: The control unit 11 outputs the objective parameter predicted in step S102 as a prediction result from the output unit 14.

[0037] In this embodiment, the explanatory factors include multiple feature quantities obtained by clustering analysis of time-series data from multiple measuring instruments during the dehydration and heating process. Figure 3 shows a conceptual diagram of the polycondensation reaction process. As shown in Figure 3, the polycondensation reaction includes a dehydration and heating process 410, a holding process 420, and a cooling process 430. Graph 401 shows the temperature transition of the material being synthesized. During the dehydration and heating process 410, the temperature of the material being synthesized increases. During the holding process 420, the temperature of the material being synthesized is maintained constant. At each intermediate stage 421-424 of the holding process 420, the quality values ​​of the material being reacted are sampled and manually analyzed. These quality values ​​include at least one of viscosity and acid value. The quality values ​​may also include physical property values ​​related to hydroxyl value and color. The duration of the holding process 420 is adjusted by performing sampling and manual analysis at each intermediate stage 421-424. At the final stage 431 of the cooling process 430, the quality values ​​of the material are analyzed.

[0038] The above analytical values ​​in the polycondensation reaction correspond to the objective factors in this embodiment. Furthermore, these analytical values ​​depend on the dehydration and heating process. On the other hand, the dehydration and heating process involves a wide variety of time-series data, so it is not realistic to use all of this time-series data as explanatory variables. Here, the time-series data includes measurements at intervals of 1 second to 1 minute. Therefore, this embodiment is characterized in that multiple feature quantities obtained by clustering analysis of time-series data from multiple measuring instruments in the dehydration and heating process are used as explanatory factors.

[0039] 4A and 4B illustrate a method for calculating multiple feature quantities for a certain lot (lot number L001). Item 501 in FIG. 4A shows the time progression of data related to the dehydration heating process for that lot. This data includes data related to the kettle temperature, degassing, column top temperature, reflux rate, partial condenser inlet / outlet temperatures, heat transfer medium, column pressure, and vapor phase temperature. In FIG. 4, data related to the dehydration heating process are represented by Data A-Data J. Data A is data related to the kettle temperature gradient (°C / min). Data B is data related to the degassing rate gradient (kg / h / min). Data C is data related to the column top temperature (°C). Data D is data related to the reflux rate gradient (kg / h / min). Data E is data related to the inlet temperature to the partial condenser (°C). Data F is data related to the outlet temperature of the partial condenser (°C). Data G is data related to the inlet temperature gradient to the heat transfer medium (°C / min). Data H is data relating to the gradient of the return temperature to the heat transfer medium (°C / min). Data I is data relating to the vessel pressure (MPa). Data J is data relating to the gradient of the vapor-phase temperature (°C / min). Each data item in item 501 is normalized. Item 503 in FIG. 4B shows the time progression of categories related to the dehydration and heating process. In this embodiment, there are five categories ranging from 0 to 4. Specifically, the categories are determined by clustering analysis of time-series data from multiple measuring instruments in the dehydration and heating process. Based on the categories, feature quantities for the dehydration and heating process are determined. For example, feature quantities are determined based on the time proportion of each category. The time proportion of each category is the cumulative time spent in each category divided by the total time. In this manner, in this embodiment, the dehydration and heating process is characterized by clustering analysis.

[0040] As described above, according to this embodiment, the multiple explanatory factors include multiple feature quantities obtained by clustering analysis of time-series data from multiple measuring instruments in the dehydration and heating process. The objective factor is characterized by including either viscosity or acid value. When predicting a polycondensation reaction, the accuracy of the prediction model can be improved by including multiple feature quantities obtained by clustering analysis of the dehydration and heating process in the explanatory factors. Therefore, this embodiment can improve prediction techniques for polycondensation reactions.

[0041] 5 and 6 show an example of the accuracy verification results of the prediction model according to this embodiment. FIG. 5 is a graph showing the viscosity of a certain polyester lot (lot number L001) predicted by the prediction model versus the actual measured value. As shown in FIG. 5, the viscosity predicted by the prediction model generally matches the actual measured value. Furthermore, the viscosity prediction accuracy is ±1.85% of the specified value. FIG. 6 is a graph showing the acid value of the above polyester lot predicted by the prediction model versus the actual measured value. As shown in FIG. 6, the acid value predicted by the prediction model generally matches the actual measured value. Furthermore, the acid value prediction accuracy is 0.085 in absolute value, relative to the specified value of 0.2 or less. Therefore, it can be seen that the accuracy of the prediction model is sufficiently high.

[0042] Here, the explanatory factors may include theoretical values ​​in a reaction physics model. In this way, the reaction physics model may be used as an explanatory factor to serve as a baseline for reaction characteristics. Similarly, the explanatory factors may include the heat transfer medium return temperature, the kettle temperature, the nitrogen form, the yield, and the nitrogen amount.

[0043] In this embodiment, a visualized graph may be output by predicting the change in the objective parameter over time during the polycondensation reaction. Specifically, at the end of the dehydration / heating process 410 (the end point 411 of the dehydration / heating process), the continuous reaction progress up to the reaction endpoint (before cooling) is predicted. Specifically, in this case, the multiple explanatory factors include a cumulative calculated value of the amount of added raw material during the polycondensation reaction (a cumulative calculated value of added raw material) and the raw material addition time. By varying the cumulative calculated value of added raw material and its addition time, which are part of the calculation conditions for the predicted change in the objective parameter over time, a visualized graph showing the relationship between the reaction end time, the amount of added raw material, and the raw material addition time is output. Here, the reaction end time refers to the time required for the material to reach the target physical property value. The amount of added raw material refers to the amount of added raw material required to meet the product specifications with a single adjustment charge.

[0044] Such a visualized graph is any graph in which the first axis represents the time of adding raw materials and the second axis represents the amount of raw materials added. For example, visualized graphs include heat maps, contour maps, etc. An example of a visualized graph is shown in FIG. 7. FIG. 7 is an example of a visualized graph where the visualized graph is a heat map. In the heat map of FIG. 7, the horizontal axis corresponds to the first axis and represents the time of adding raw materials. The vertical axis of the heat map corresponds to the second axis and represents the amount of raw materials added. A numerical value for the reaction completion time may be displayed in each cell. In such a heat map, the shade of each cell changes depending on the reaction completion time. Specifically, the shorter the reaction completion time, the higher the concentration of the cell. This visualizes the relationship between the addition time and amount of raw materials added and the reaction completion time. In other words, a visualized graph such as a heat map makes it easy to visually grasp the addition time and amount of raw materials added to achieve the shortest reaction completion time.

[0045] Here, the visualized graph may include a plot showing performance data. Plot 801 in Fig. 7 is performance data in which the amount of additional raw material added is 7 and the addition time is 170. By comparing the performance data with the performance data, it is possible to determine the optimal time and amount of additional raw material to be added.

[0046] The prediction model according to this embodiment may be, for example, a neural network model. When the prediction model is a neural network model, the coefficients of the activation function of the neural network model may be different between the intermediate layer and the output layer. For example, the coefficients of the activation function for the intermediate layer are larger than the coefficients of the activation function for the output layer.

[0047] FIG. 8 shows a conceptual diagram of a neural network model according to this embodiment. The neural network model includes an input layer 100, an intermediate layer 200, and an output layer 300. The neural network model according to this embodiment is fully connected. In this embodiment, the number of layers in the neural network model is, for example, two. This number of layers is the number of layers excluding the input layer. By setting the number of layers in the neural network model to two, it is possible to prevent a model shape that is incompatible with the physical phenomena occurring in a polycondensation reaction. In other words, by minimizing the number of layers in the neural network model, it is possible to realize a model shape that is suitable for the physical phenomena occurring in a polycondensation reaction. The number of layers in the neural network model according to this embodiment is not limited to three layers and may be three or more. When the number of layers in the neural network model is three or more, the activation function coefficient may be set to be larger toward the earlier layers of the neural network model.

[0048] The input layer 100 includes a plurality of elements 101-104 (also referred to as input elements 101-104). In the neural network model shown in FIG. 8, the number of input elements is four. The input elements 101-104 are also referred to as the first to fourth elements, respectively. Explanatory factors are input to each of the input elements 101-104. Note that the number of input elements is not limited to this, and may be less than four or five or more.

[0049] The hidden layer 200 includes a plurality of elements 201-214 (also referred to as middle elements 201-214). In the neural network model shown in FIG. 8, there are 14 middle elements. The middle elements 201-214 are also referred to as the 1st to 14th elements, respectively. Note that the number of middle elements is not limited to this, and may be less than 14 or 15 or more.

[0050] The output layer 300 includes an element 301 (output element 301). In the neural network model shown in FIG. 8, the number of output elements is one. The output element 301 is also called the first element. Note that the number of output elements is not limited to this, and may be two or more.

[0051] The values ​​input from the input elements 101-104 of the input layer 100 to the intermediate elements 201-214 of the intermediate layer 200 are transformed in the intermediate layer 200 based on the activation function of the intermediate layer 200. The transformed values ​​are output to the element 301 of the output layer 300. The activation function of the intermediate layer 200 is, for example, a sigmoid function. The values ​​input from the intermediate elements 201-214 of the intermediate layer 200 to the output element 301 of the output layer 300 are transformed in the output layer 300 based on the activation function of the output layer 300 and output. The activation function of the output layer 300 is, for example, a sigmoid function. Specifically, the activation functions of the intermediate layer and the output layer are, for example, sigmoid functions defined by the following mathematical expressions (1) and (2), respectively.

[0052]

[0053] In the neural network model according to this embodiment, the coefficients of the activation function for the intermediate layer are larger than the coefficients of the activation function for the output layer. This allows the configuration of the neural network model to be optimized when making predictions regarding polycondensation reactions. Specifically, in a neural network model that makes predictions regarding polycondensation reactions, it is desirable that changes in explanatory factors be perceived as clear changes. Therefore, by making the coefficients of the activation function for the intermediate layer larger than the coefficients of the activation function for the output layer, changes in input values ​​to the intermediate layer can be transmitted to the output layer as clear changes. On the other hand, in the output layer of a neural network model that makes predictions regarding polycondensation reactions, it is necessary to converge the values ​​of the training data and the objective factor. Therefore, the coefficients of the activation function for the output layer are set smaller than the coefficients of the activation function for the intermediate layer. In this way, the value of the objective factor output from the output layer is fine-tuned.

[0054] Furthermore, by making the coefficients of the activation function different between the intermediate layer and the output layer, the learning process of the neural network model can be optimized. Specifically, by changing the coefficients of the activation function, it is possible to adjust the amount of update of the weight variables in the output layer and intermediate layer during the learning process. Furthermore, updating the weight variables has a significant impact on the learning process. Therefore, the learning process can be optimized based on adjusting the amount of update.

[0055] Specifically, in a neural network model for making predictions regarding polycondensation reactions, it is preferable to make the amount of update of the weight variables related to the intermediate layer relatively large. This allows the weight variables in the intermediate layer to fluctuate more significantly during the learning process, and changes in the input values ​​to the intermediate layer can be transmitted to the output layer as clear changes. On the other hand, it is preferable to make the amount of update of the weight variables related to the output layer relatively small. This allows the weight variables in the output layer to fluctuate less significantly during the learning process, and the values ​​of the training data and the objective factor can more easily converge. In addition, a l >a L By satisfying this condition, it becomes possible to approximate any smooth function with sufficient accuracy, eliminating the need to unnecessarily increase the number of hidden layers. This means that sufficient accuracy can be obtained even with a single hidden layer. Having fewer hidden layers directly leads to suppressing overfitting, which has the secondary effect of improving the stability of the learning process and the robustness of the model.

[0056] In this embodiment, the activation functions of the intermediate and output layers are sigmoid functions, but the activation functions are not limited to sigmoid functions. For example, the activation functions of the intermediate and output layers may be functions such as hyperbolic tangent functions (tanh functions) and ramp functions (ReLU).

[0057] Although the present disclosure has been described based on the drawings and examples, it should be noted that those skilled in the art can easily 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 means or step can be rearranged so as not to be logically inconsistent, and multiple means or steps can be combined or divided into one.

[0058] 10 Information processing device 11 Control unit 12 Memory unit 13 Input unit 14 Output unit 100 Input layer 200 Intermediate layer 300 Output layer 101-104, 201-214, 301 Element 401 Graph 410 Dehydration and temperature increase process 411 End point of dehydration and temperature increase process 420 Hold process 421-424 Intermediate stage 430 Cooling process 431 Final stage 501, 503 Item 801 Plot

Claims

1. A method for making predictions regarding a polycondensation reaction, executed by an information processing device, comprising: training a predictive model based on performance data including a plurality of explanatory factors and a target factor related to the polycondensation reaction; predicting the objective factor during the polycondensation reaction based on a plurality of explanatory factors related to the polycondensation reaction by the prediction model; Including, the plurality of explanatory factors include a plurality of feature quantities determined based on a time ratio of a category related to the dehydration and heating process obtained by performing a clustering analysis on time-series data of a plurality of measuring instruments in the dehydration and heating process, The method, wherein the objective factors include at least one of viscosity and acid number.

2. 2. The method of claim 1 , The method, wherein the plurality of explanatory factors comprises theoretical values ​​in a reaction physics model.

3. The method according to claim 1 , wherein the polycondensation reaction is a dehydration condensation reaction of a polyester.

4. 2. The method of claim 1 , In the step of predicting, a change in the target factor over time during the polycondensation reaction is predicted in advance; The plurality of explanatory factors include a raw material addition cumulative calculation value and a raw material addition time during the polycondensation reaction, The method further comprises: A method for outputting a visualized graph showing the relationship between the reaction end time, the amount of raw material added, and the raw material addition time, by varying a raw material addition cumulative calculation value and raw material addition time, which are part of the calculation conditions for the predicted value of the time-dependent change of the objective parameter,

5. 5. The method of claim 4, The method, wherein the visualization graph is either a heat map or a contour map with a first axis being ingredient addition time and a second axis being ingredient addition amount.

6. 5. The method of claim 4, The method, wherein the visualization graph includes a plot showing performance data.

7. 2. The method of claim 1 , The method, wherein the predictive model is a neural network model including an input layer, an intermediate layer, and an output layer, and a coefficient of an activation function associated with the intermediate layer is greater than a coefficient of an activation function associated with the output layer.

8. An information processing device comprising a control unit and performing predictions related to a polycondensation reaction, The control unit is training a predictive model based on performance data including a plurality of explanatory factors and a target factor related to the polycondensation reaction; predicting the objective factor during the polycondensation reaction based on a plurality of explanatory factors related to the polycondensation reaction using the prediction model; the plurality of explanatory factors include a plurality of feature quantities determined based on a time ratio of a category related to the dehydration and heating process obtained by performing a clustering analysis on time-series data of a plurality of measuring instruments in the dehydration and heating process, The objective factor includes any one of viscosity and acid value.

9. A program for making predictions regarding polycondensation reactions executed by an information processing device, comprising: training a predictive model based on performance data including a plurality of explanatory factors and a target factor related to the polycondensation reaction; predicting the objective factor during the polycondensation reaction based on a plurality of explanatory factors related to the polycondensation reaction by the prediction model; Run the command, the plurality of explanatory factors include a plurality of feature quantities determined based on a time ratio of a category related to the dehydration and heating process obtained by performing a clustering analysis on time-series data of a plurality of measuring instruments in the dehydration and heating process, The program, wherein the objective factor includes any one of viscosity and acid number.