Method and apparatus for predicting the durability properties of resin compositions

Abstract

To provide a method and a device for predicting durability physical property of a resin composition which can accurately predict an elongation residual rate and a tensile strength residual rate.SOLUTION: A method for predicting durability physical property that is physical property after a predetermined durability test in a resin composition formed using a material containing a base polymer and a filler, in which the durability physical property is an elongation residual rate or a tensile strength residual rate represented by the following expressions: elongation residual rate=100×(elongation after durability test / initial elongation), and tensile strength residual rate=100×(tensile strength after durability test / initial tensile strength), performs machine learning with respective blending amounts of at least the base polymer and the filler as explanatory variables, and the elongation residual rate or the tensile strength residual rate as an objective variable and creates a regression model 7, and predicts an elongation residual rate or a tensile strength residual rate of a resin composition of a prediction object by using the regression model 7.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a method and apparatus for predicting the durability physical properties of a resin composition.

Background Art

[0002] In recent years, methods and apparatuses for designing materials using the learning results of machine learning have been proposed. By using such methods and apparatuses, for example, it becomes possible to predict physical properties according to the blending amount of raw materials, etc., shorten the development period of the material, and suppress the development cost.

[0003] Note that as prior art document information related to the invention of this application, there is Patent Document 1.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] By the way, when designing a resin composition used as a coating material for electric wires, it is desirable to accurately predict the durability physical properties. When evaluating the deterioration of physical properties by durability tests such as heat resistance tests and oil resistance tests, it is necessary to evaluate how much the physical properties after the durability test have decreased with respect to the physical properties in the initial state before the durability test. As the durability physical properties used as such an evaluation index, the elongation retention rate, which is the change rate of elongation before and after the durability test, and the tensile strength retention rate, which is the change rate of tensile strength before and after the durability test, are generally used, and it is desirable to accurately predict these elongation retention rate and tensile strength retention rate.

[0006] Therefore, an object of the present invention is to provide a method and apparatus for predicting the durability physical properties of a resin composition capable of accurately predicting the elongation retention rate and the tensile strength retention rate. [Means for solving the problem]

[0007] The present invention aims to solve the above problems and provides a method for predicting the durable physical properties of a resin composition formed using a material comprising a base polymer and a filler, wherein the durable physical properties are the elongation retention rate or tensile strength retention rate shown in the following formula, Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) The present invention provides a method for predicting the durability properties of a resin composition, which involves creating a regression model by performing machine learning with at least the amounts of the base polymer and the filler used as explanatory variables and the elongation retention rate or tensile strength retention rate as the objective variable, and then using the regression model to predict the elongation retention rate or tensile strength retention rate of the resin composition to be predicted.

[0008] Furthermore, the present invention aims to solve the above problems and provides a method for predicting the durable physical properties of a resin composition formed using a material comprising a base polymer and a filler, wherein the durable physical properties are the elongation retention rate or tensile strength retention rate shown in the following formula, Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) At least the respective amounts of the base polymer and the filler and the following (a) (a) A volume fraction parameter including information on the ratio of the volume of the base polymer to the volume of the filler. The present invention provides a method for predicting the durability properties of a resin composition, which involves creating a regression model by machine learning using synthetic characteristic values ​​including as explanatory variables and elongation or tensile strength after the durability test as the dependent variable, predicting the elongation or tensile strength of the target resin composition after the durability test using the regression model, and predicting the remaining elongation rate or tensile strength rate of the target resin composition based on the predicted elongation or tensile strength after the durability test.

[0009] Furthermore, the present invention aims to solve the above problems and provides an apparatus for predicting the durable physical properties of a resin composition formed using a material comprising a base polymer and a filler, wherein the durable physical properties are the elongation retention rate or tensile strength retention rate shown in the following formula, Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) The present invention provides a resin composition durability property prediction device, comprising: a regression model creation processing unit that creates a regression model by performing machine learning with at least the amounts of the base polymer and the filler used as explanatory variables and the elongation retention rate or tensile strength retention rate as the objective variable; and a durability property prediction processing unit that uses the regression model to predict the elongation retention rate or tensile strength retention rate of the resin composition to be predicted.

[0010] Furthermore, the present invention aims to solve the above problems and provides an apparatus for predicting the durable physical properties of a resin composition formed using a material comprising a base polymer and a filler, wherein the durable physical properties are the elongation retention rate or tensile strength retention rate shown in the following formula, Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) At least the respective amounts of the base polymer and the filler and the following (a) (a) A volume fraction parameter including information on the ratio of the volume of the base polymer to the volume of the filler. The present invention provides a resin composition durability property prediction device, comprising: a regression model creation processing unit for creating a regression model for predicting post-test physical properties using synthetic characteristic values ​​including as explanatory variables and elongation or tensile strength after the durability test as the objective variable by performing machine learning; a post-test physical property prediction processing unit that predicts the elongation or tensile strength of the target resin composition after the durability test using the post-test physical property prediction regression model; and a durability property prediction processing unit that predicts the elongation retention rate or tensile strength retention rate of the target resin composition based on the elongation or tensile strength after the durability test predicted by the post-test property prediction processing unit.

Advantages of the Invention

[0011] According to the present invention, it is possible to provide a method and an apparatus for predicting the durability physical properties of a resin composition that can accurately predict the elongation residual rate and the tensile strength residual rate.

Brief Description of the Drawings

[0012] [Figure 1] It is a schematic configuration diagram of a durability physical property prediction apparatus for a resin composition according to the first embodiment of the present invention. [Figure 2] It is a diagram showing an example of learning data. [Figure 3] (a) is a diagram for explaining the regression model creation process, and (b) is a diagram for explaining the durability physical property prediction process. [Figure 4] It is a flowchart of a method for predicting the durability physical properties of a resin composition according to the first embodiment of the present invention. [Figure 5] (a) is a flowchart of data acquisition processing, and (b) is a flowchart of regression model creation processing. [Figure 6] It is a flowchart of durability physical property prediction processing. [Figure 7] It is a diagram showing the examination results of explanatory variables suitable in the first embodiment. [Figure 8] (a) is a diagram showing the examination results of explanatory variables suitable in the heat resistance test, and (b) is a diagram showing the examination results of explanatory variables suitable in the oil resistance test. [Figure 9] (a) is a diagram showing the examination results of explanatory variables suitable in the fuel resistance test, and (b) is a diagram showing the examination results of explanatory variables suitable in the low temperature resistance test. [Figure 10] It is a schematic configuration diagram of a durability physical property prediction apparatus for a resin composition according to the second embodiment of the present invention. [Figure 11] (a) is a diagram for explaining the regression model creation process for predicting physical properties after testing, and (b) is a diagram for explaining the physical property prediction process after testing. [Figure 12] (a) is a diagram for explaining the regression model creation process for predicting initial physical properties, and (b) is a diagram for explaining the initial physical property prediction process. [Figure 13] It is a diagram for explaining the prediction accuracy of the second embodiment. [Modes for carrying out the invention]

[0013] [Embodiment] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

[0014] (First Embodiment) Figure 1 is a schematic diagram of the durability property prediction device 1 for resin compositions according to the first embodiment (hereinafter simply referred to as the durability property prediction device 1). The durability property prediction device 1 is a device that predicts the durability property of a resin composition formed using a material containing a base polymer and a filler.

[0015] (Regarding predicted durability) Durability refers to the physical properties after a predetermined durability test. In this embodiment, the durability is predicted as the elongation retention rate or tensile strength retention rate shown in the following formulas (1) and (2). Elongation rate = 100 × (Elongation after durability test / Initial elongation) ... (1) Tensile strength remaining percentage = 100 × (Tensile strength after durability test / Initial tensile strength) ... (2)

[0016] The inventors investigated methods for predicting elongation and tensile strength retention rates and found that a method that directly predicts elongation and tensile strength retention rates using machine learning (referred to as the direct method) can improve prediction accuracy compared to a method that predicts elongation and tensile strength after durability testing using machine learning and then calculates the elongation and tensile strength retention rates using equations (1) and (2) above based on the prediction results (referred to as the indirect method). Furthermore, they found that the prediction accuracy of the indirect method can be improved by including the composite characteristic values ​​described later as explanatory variables. The direct method will be described in the first embodiment. The indirect method will be described in the second embodiment described later.

[0017] (Outline configuration of the durability property prediction device 1) As shown in Figure 1, the durability prediction device 1 includes a control unit 2, a storage unit 3, a display unit 4, and an input device 5. In this embodiment, the durability prediction device 1 is configured using a personal computer.

[0018] The control unit 2 is implemented by appropriately combining computing elements such as a CPU, memory, interfaces, software, storage devices, etc. In this embodiment, the control unit 2 includes a data acquisition processing unit 21, a regression model creation processing unit 22, a durability property prediction processing unit 23, and a predicted durability property presentation processing unit 24. Details of each part will be described later.

[0019] The memory unit 3 is implemented by a predetermined storage area of ​​memory or a storage device. The display unit 4 is, for example, a liquid crystal display, and the input device 5 is, for example, a keyboard or mouse. The display unit 4 may be configured as a touch panel, and the display unit 4 may also serve as the input device 5. Furthermore, the display unit 4 and the input device 5 may be configured separately from the durability property prediction device 1 and be able to communicate with each other via wireless communication or the like. In this case, the display unit 4 or the input device 5 may be a mobile terminal such as a tablet or smartphone.

[0020] (Regarding composite property values) As will be described in detail later, in the durability property prediction device 1 according to this embodiment, a regression model 7 is created by machine learning using at least the amounts of base polymer and filler as explanatory variables and the elongation retention rate or tensile strength retention rate as the objective variable, and the elongation retention rate or tensile strength retention rate of the resin composition to be predicted is used with this regression model 7. At this time, further accuracy can be improved by adding a composite characteristic value, which is a combination of characteristic values ​​related to elongation and tensile strength for the entire resin composition, as an explanatory variable.

[0021] The following parameters (a) to (f) can be used as composite characteristic values. When using composite characteristic values ​​as explanatory variables, it is desirable to include at least the volume fraction parameter (a). Furthermore, in addition to the volume fraction parameter (a), parameters (b) to (f) may be added as explanatory variables as appropriate. In this embodiment, all of these composite characteristic values ​​(a) to (f) were used as explanatory variables. The details of each parameter are described below.

[0022] (a) Volume fraction parameter The volume fraction parameter is a parameter that contains information about the ratio of the volume of the base polymer to the volume of the filler. The filler volume fraction and base polymer volume fraction, calculated using the following formula, can be used as the volume fraction parameter. Note that "total volume of materials" in the formula refers to the total volume of the resin composition. (Filler volume fraction) = (Filler volume) / (Total material volume) × 100 (Base polymer volume fraction) = (Base polymer volume) / (Total material volume) × 100 For example, the base polymer volume is calculated by dividing the amount of base polymer used (g) by the density of the base polymer (g / cm³). 3 It can be found by dividing by ).

[0023] Furthermore, instead of the filler volume fraction and base polymer volume fraction mentioned above, the filler volume ratio calculated by the following formula can also be used as the volume fraction parameter. (Filler volume ratio) = (Filler volume ratio) / (Base polymer volume ratio) By using the filler volume ratio as the volume fraction parameter, it becomes possible to reduce the number of explanatory variables, thereby lowering the computational load during machine learning.

[0024] (b) Filler surface area The filler surface area is a parameter that includes information on the BET (Brunauer-Emmett-Teller) specific surface area of ​​the filler, and is used as a parameter that represents the particle size of fillers used as flame retardants or flame retardant additives. This is because the particle size of the filler is thought to affect the elongation and tensile strength of the resin composition. In this embodiment, the filler surface area used as filler surface area data was calculated using the following formula. (Filler surface area) = (BET specific surface area) × (Amount of filler) In the above formula, the BET specific surface area is the specific surface area measured by the BET method. The amount of filler is the total amount (parts by mass) of filler, including the flame retardant and flame retardant additive.

[0025] (c) Amount of maleic anhydride modification The amount of maleic anhydride modification is a parameter that represents the amount (proportion) of maleic anhydride (MAH) contained in the resin composition. Since maleic anhydride plays a role in bonding the polymer and filler, it is thought that the amount (proportion) of maleic anhydride affects the elongation and tensile strength of the resin composition. In this embodiment, the amount of maleic anhydride modification was determined by the following formula. (Degree of maleic anhydride modification) = (Degree of modification of maleic anhydride material) × (Amount of maleic anhydride material used) / 100 In the above formula, "maleic anhydride material" refers to a base polymer that contains maleic anhydride (i.e., a polymer modified with maleic anhydride).

[0026] (d) Amount of crystals The amount of crystals is a parameter that represents the ratio of the amount of crystalline polymer (amount in the blend) to the amount of base polymer (amount in the blend). In this embodiment, the amount of crystals was determined by the following formula. (Amount of crystals) = (Amount of crystalline polymer) / (Amount of total material) × 100

[0027] (e) Copolymerization amount The copolymerization amount is a parameter that represents the proportion of the copolymer component with ethylene (component copolymerized with ethylene) contained in the base polymer. Examples of copolymer components with ethylene include vinyl acetate groups. In this embodiment, the copolymerization amount was determined by the following formula. (Copolymerization amount) = (Amount of copolymer component with ethylene contained in the base polymer) / (Total amount of base polymer) × 100 The amount of copolymer component with ethylene contained in the base polymer in the above formula can be determined by multiplying the amount of each polymer contained in the base polymer by the amount of copolymer component with ethylene, and then summing up the resulting values ​​for all polymers.

[0028] (f) Filler identification information (filler category information) Filler identification information is a categorization parameter that represents category information such as the type of flame retardant contained in the filler and information on the surface treatment method of the flame retardant. In this embodiment, a total of four parameters were used as filler identification information: a parameter indicating whether or not the flame retardant is surface-treated with a fatty acid, a parameter indicating whether or not the flame retardant is surface-treated with a silane, a parameter indicating whether or not the flame retardant contains magnesium, and a parameter indicating whether or not the flame retardant contains aluminum. Each parameter of the filler identification information is represented by either "0" (does not contain, not applicable) or "1" (contains, applicable) (see Figure 2).

[0029] (Regarding training data 6) Next, we will describe the training data 6 used for machine learning. Figure 2 is a diagram showing an example of training data 6. Note that Figure 2 is a conceptual representation of training data 6 and does not contain actual experimental data, etc. As shown in Figure 2, training data 6 is a database that includes at least blending amount data 11 containing information on the blending amounts of each material, synthesis property value data 12 containing the information described in (a) to (f) above, process data 13 which are the manufacturing conditions of the resin composition, initial property data 14 which are information on initial physical properties (initial elongation and tensile strength), and durability property data 15 which are information on durability properties after durability testing. In the illustrated example, training data 6 includes ID data (ID) 10 for identifying the resin composition. Note that training data 6 may also include other information, for example, process data 13 may further include data representing other manufacturing conditions, or data representing the microstructure of the resin composition (so-called microstructure data), etc.

[0030] The blending amount data 11 includes information on the blending amounts of each material constituting the resin composition. More specifically, the blending amount data 11 includes information on the blending amounts of the base polymer, filler, and other materials. The synthetic property value data 12 includes the above information (a) to (f), namely (a) volume fraction parameter (here, filler volume ratio), (b) filler surface area, (c) amount of maleic anhydride modification, (d) amount of crystals, (e) amount of copolymerization, and (f) filler identification information. Note that the synthetic property value data 12 may include data other than the above six data, and may not include some of the above six data.

[0031] In this embodiment, since the resin composition using electron beam irradiation for crosslinking is the target of property prediction, the training data 6 includes irradiation dose data as process data 13, which includes information on the amount of electron beam irradiation during crosslinking. Note that if the target is a resin composition that does not undergo electron beam irradiation for crosslinking, the irradiation dose data is omitted. In this embodiment, initial property data 14 is included to further improve the prediction accuracy of elongation residual rate and tensile strength residual rate, but initial property data 14 can be omitted. Durability property data 15 includes information on the elongation residual rate and tensile strength residual rate to be predicted, as well as information on elongation and tensile strength after the durability test. Of the durability property data 15, the data showing information on elongation and tensile strength after the durability test is called post-test property data 16. In this embodiment, post-test property data 16 is not used and can be omitted. However, in the second embodiment described later (see Figure 8), post-test property data 16 is essential.

[0032] (Regarding durability testing) In this embodiment, three durability tests were conducted: a heat resistance test, an oil resistance test, a fuel resistance test, and a low-temperature resistance test. The shape of the test specimen was a predetermined dumbbell shape in all durability tests.

[0033] In the heat resistance test, a dumbbell-shaped test specimen was placed in a gear oven and heated at 135°C for 168 hours, then left to stand at room temperature for one day, and a tensile test was performed at room temperature. In the oil resistance test, a dumbbell-shaped test specimen was immersed in 100°C oil for 72 hours, then removed from the oil and left to stand at room temperature for one day, and a tensile test was performed at room temperature.

[0034] In the fuel resistance test, a dumbbell-shaped test specimen was immersed in fuel at 70°C for 168 hours, then removed from the fuel and left to stand at room temperature for one day before undergoing a tensile test at room temperature. In the low-temperature resistance test, a dumbbell-shaped test specimen was subjected to a tensile test at -40°C. In the low-temperature resistance test, only elongation was tested.

[0035] (Data acquisition processing unit 21) The data acquisition processing unit 21 performs data acquisition processing (see Figure 5(a)) to store the training data 6 input from an external source in the storage unit 3. The training data 6 may be input by the input device 5, by communication from an external device (wired communication, wireless communication, or communication via a network, etc.), or by media such as a USB memory. Furthermore, the data acquisition processing unit 21 may be configured to actively acquire the training data 6, such as by sending a signal to the external device requesting the training data 6.

[0036] (Regression model creation processing unit 22) The regression model creation processing unit 22 performs machine learning using the training data 6 acquired by the data acquisition processing unit 21 and creates a regression model 7 (see Figure 5(b)). As shown in Figure 3(a), the regression model creation processing unit 22 receives the following as training data 6: blending amount data 11, which is data on the blending amounts; composite characteristic value data 12, which is data on each composite characteristic value; process data 13, which includes irradiation amount data; initial physical property data 14, which is data on the initial elongation or tensile strength; and durability physical property data 15, which is data on the elongation retention rate or tensile strength retention rate. Of these, the blending amount, composite characteristic value, irradiation amount, and initial physical property (initial elongation or tensile strength) are used as explanatory variables, and the durability physical property (elongation retention rate or tensile strength retention rate) is used as the dependent variable.

[0037] Note that it is not mandatory to use composite characteristic values, process data (irradiation dose), and initial physical properties as explanatory variables. However, using these composite characteristic values, process data (irradiation dose), and initial physical properties as explanatory variables makes it possible to predict durability properties (elongation retention rate or tensile strength retention rate) with greater accuracy. Furthermore, for composite characteristic values, only the parameters selected from the data (a) to (f) above, according to the durability property to be predicted, may be used as explanatory variables. Also, when predicting elongation retention rate, it is preferable to use the initial elongation as the initial physical property, and when predicting tensile strength retention rate, it is preferable to use the initial tensile strength as the initial physical property.

[0038] The regression model creation processing unit 22 includes software such as a learning algorithm for learning, using machine learning, the correlation between each parameter used as an explanatory variable and the target variable, the durable physical property (elongation retention rate or tensile strength retention rate), from the input training data 6. The learning algorithm is not particularly limited, and any known learning algorithm can be used, for example, so-called deep learning using a neural network with three or more layers can be used. What the regression model creation processing unit 22 learns corresponds to a model structure that represents the correlation between the explanatory variables and the target variable.

[0039] The regression model creation processing unit 22 creates a regression model 7 that represents the correlation between explanatory variables (composition amount, composite characteristic value, irradiation dose, and initial physical properties) and the objective variable (remaining elongation or remaining tensile strength) based on the input training data 6.

[0040] More specifically, the regression model creation processing unit 22 iteratively performs learning based on the input training data 6, using a data set that includes explanatory variables (mixture amount, composite characteristic value, irradiation dose, and initial physical properties) and a target variable (remaining elongation or remaining tensile strength), and automatically interprets the correlation between the two. At the start of learning, the correlation is unknown, but as learning progresses, the correlation between the explanatory variables (mixture amount, composite characteristic value, irradiation dose, and initial physical properties) and the target variable (remaining elongation or remaining tensile strength) is gradually interpreted, and by using the resulting trained regression model 7, it becomes possible to interpret the correlation between the explanatory variables (mixture amount, composite characteristic value, irradiation dose, and initial physical properties) and the target variable (remaining elongation or remaining tensile strength).

[0041] The regression model creation processing unit 22 stores the created regression model 7 in the storage unit 3. In this embodiment, the regression model creation processing unit 22 updates the regression model 7 each time the training data 6 is updated. However, this is not the only method; for example, when performing the durability property prediction process described later, the updated training data 6 may be learned all at once, and the regression model 7 may be updated.

[0042] In this embodiment, since the resin composition using crosslinking by electron beam irradiation is the target of property prediction, irradiation dose data is used as an explanatory variable. However, if the resin composition does not undergo crosslinking by electron beam irradiation, the irradiation dose data can be omitted. In addition, other data (process data, tissue data, etc.) may be added to the explanatory variables as needed.

[0043] (Durability Prediction Processing Unit 23) The durability property prediction processing unit 23 uses a regression model 7 to perform durability property prediction processing (see Figure 6) to predict the elongation retention rate or tensile strength retention rate of the resin composition to be predicted.

[0044] As shown in Figure 3(b), in the durability property prediction process, the regression model 7 and the prediction source data 8 are input to the durability property prediction processing unit 23. The same parameters used as explanatory variables when creating the regression model 7 are used for the prediction source data 8. Hereinafter, for the sake of distinction, the blending amount data 11 used as the prediction source data 8 will be referred to as prediction source blending amount data 11a, the composite characteristic value data 12 used as the prediction source data 8 will be referred to as prediction source composite characteristic value data 12a, the process data 13 used as the prediction source data 8 will be referred to as prediction source process data 13a, and the initial property data 14 used as the prediction source data 8 will be referred to as prediction source initial property data 14a.

[0045] The durability property prediction processing unit 23 uses a regression model 7 to obtain durability property data 15 corresponding to the prediction source data 8, and stores the obtained durability property data 15 as prediction data 9 in the storage unit 3. Hereinafter, the durability property data 15 predicted by the durability property prediction processing will be referred to as prediction durability property data 15a. The prediction data 9 obtained here represents the durability property (elongation retention rate or tensile strength retention rate) of the resin composition predicted when using the materials etc. set in the prediction source data 8.

[0046] (Predictive durability property display processing unit 24) The predictive durability property presentation processing unit 24 performs predictive durability property presentation processing to present the predicted data 9. In the predictive durability property presentation processing, for example, the predicted data 9 is displayed on the display unit 4. The predictive durability property presentation processing may also be configured to present data other than the predicted data 9, for example, items of the composite characteristic value data 12 used for prediction.

[0047] (Method for predicting the durability of resin compositions) (Main routine) Figure 4 is a flowchart of the method for predicting the durability properties of a resin composition according to this embodiment. In Figure 4, solid arrows represent the control flow, and dashed arrows represent the input and output of signals and data.

[0048] As shown in Figure 4, in the method for predicting the durability of a resin composition according to this embodiment, first, in step S1, the control unit 2 determines whether new data has been input. If NO (N) is determined in step S1, the process proceeds to step S5. If YES (Y) is determined in step S1, the process proceeds to step S2, where data acquisition processing is performed.

[0049] In the data acquisition process of step S2, as shown in Figure 5(a), the data acquisition processing unit 21 receives the blending amount data 11, the composite characteristic value data 12, the process data 13, the initial physical property data 14, and the durability physical property data 15 (step S21), links the received data together, and stores them in the storage unit 3 as learning data 6 (step S22). After that, it returns.

[0050] After the data acquisition process in step S2, the regression model creation process is performed in step S3. In the regression model creation process, as shown in Figure 5(b), first, in step S31, the regression model creation processing unit 22 updates the regression model 7 using the untrained training data 6 for machine learning. If the regression model 7 has not yet been created, a new regression model 7 is created in step S31. Then, in step S32, the updated (or created) regression model 7 is stored in the storage unit 3 and returned.

[0051] When predicting the physical properties of a resin composition, the source data 8 is input using the input device 5, etc. (step S4). Alternatively, the data to be used as the source data 8 (source blending amount data 11a, source synthesis characteristic value data 12a, source process data 13a, and source initial physical property data 14a) may be input into the resin composition durability property prediction device 1 in advance, and the input device 5 may be configured to select the data to be used as the source data 8.

[0052] In step S5, the control unit 2 determines whether the prediction source data 8 has been input. If the result in step S5 is NO, it returns (returns to step S1). If the result in step S5 is YES, it proceeds to step S6.

[0053] In step S6, the durability property prediction process is performed. In the durability property prediction process, as shown in Figure 6, first, in step S61, the durability property prediction processing unit 23 predicts the durability property (remaining elongation rate or remaining tensile strength rate) corresponding to the prediction source data 8 using the regression model 7, and the predicted data, the predicted durability property data 15a, is set as the prediction data 9. Then, in step S62, the obtained prediction data 9 is stored in the storage unit 3. After that, the process returns and proceeds to step S7 in Figure 4.

[0054] In step S7, the predicted durability property presentation process is performed. In the predicted durability property presentation process, the predicted data 9 predicted in step S6 is presented, for example, by displaying it on the display unit 4. After that, the process returns (returns to step S1).

[0055] (Consideration of suitable explanatory variables) Using data including the blending ratio, the composite characteristic values ​​(a) to (f) above, the electron beam irradiation dose, and the initial physical properties, regression model 7 was created by changing the parameters used as explanatory variables, and the prediction accuracy of the created regression model 7 was evaluated by cross-validation. More specifically, 70% of the total data was divided into training data 6 and the remaining 30% into test data. After creating regression model 7 using the divided training data 6, the test data was evaluated using the created regression model 7, and the coefficient of determination (R) was calculated. 2 The mean absolute error (MAE) was calculated. This data splitting, the creation of regression model 7, and evaluation were repeated 30 times, and the coefficient of determination R 2 The mean absolute error was calculated for each. 2 For the first metric, a larger value indicates higher prediction accuracy, while for the second metric, a smaller value indicates higher prediction accuracy.

[0056] In the study, the target of prediction was the elongation after the oil resistance test. Examples 1, which used only the amount of compounding as the explanatory variable, Example 2, which used the amount of compounding and the synthetic characteristic value, Example 3, which used the amount of compounding and the initial physical property (initial elongation), and Example 4, which used the amount of compounding, the synthetic characteristic value, and the initial physical property were each examined. Although not shown in the figures, process data 13 (irradiation dose) was also included as an explanatory variable in all Examples 1 to 4. The results of the study are shown in Figure 7.

[0057] As shown in Figure 7, compared to Example 1, which used only the blending amount as an explanatory variable, Examples 2 to 4, which added synthetic characteristic values ​​and initial physical properties as explanatory variables, showed improved prediction accuracy. Example 4, which added both synthetic characteristic values ​​and initial physical properties to the blending amount, showed the highest prediction accuracy. From these results, it was confirmed that when predicting elongation residual rate and tensile strength residual rate, adding synthetic characteristic values ​​and initial physical properties to the explanatory variables can improve prediction accuracy.

[0058] Furthermore, using data different from that shown in Figure 7, the prediction accuracy was evaluated for each durability test: heat resistance test (residual tensile strength), oil resistance test (residual tensile strength and elongation), fuel resistance test (residual tensile strength and elongation), and low-temperature resistance test (residual elongation at -40°C and -50°C). Here, evaluations were conducted for cases where only the blending amount was used as the explanatory variable, where only the synthetic characteristic value was used as the explanatory variable, and where both the blending amount and the synthetic characteristic value were used as the explanatory variable. In all three cases, process data 13 (irradiation dose) was also included as an explanatory variable. As for the synthetic characteristic value, all of the above parameters (a) to (f) were used, and for the volume fraction parameter (a), two parameters were used: filler volume fraction and base polymer volume fraction.

[0059] For evaluation, as in the case of Figure 7, 70% of the total data was divided into training data 6 and the remaining 30% into test data. A regression model 7 was created using the divided training data 6, and then the test data was evaluated using the durable properties (remaining elongation or remaining tensile strength) obtained using the created regression model 7, and the coefficient of determination R 2 The mean absolute error (MAE) was calculated. This data splitting, the creation of regression model 7, and evaluation were repeated 30 times, and the coefficient of determination R 2 The mean absolute error and the mean value of the mean absolute error were calculated. The results are shown in Figures 8 and 9, respectively.

[0060] As shown in Figures 8 and 9, in all durability tests, it was confirmed that using both the compounding amount and the composite characteristic value as explanatory variables improved prediction accuracy compared to using only the compounding amount. Furthermore, it was confirmed that using only the composite characteristic value as the explanatory variable often resulted in prediction accuracy equivalent to or slightly lower than that of using only the compounding amount. Additionally, it was confirmed that using both the compounding amount and the composite characteristic value tended to result in higher prediction accuracy than using only the composite characteristic value. Thus, it is possible to improve prediction accuracy by using both the compounding amount and the composite characteristic value as explanatory variables.

[0061] (modified version) In this embodiment, measured values ​​were used as the initial physical property data 14a for prediction, but the initial elongation or tensile strength values ​​predicted using machine learning may also be used as the initial physical property data 14a for prediction. In this case, separate from the regression model 7 described above, a regression model for predicting initial physical properties can be created by performing machine learning with the blending amount and synthesis characteristic values ​​as explanatory variables and the initial physical properties (initial elongation or tensile strength) as the objective variable, and the obtained regression model for predicting initial physical properties can be used to predict the initial physical properties (initial elongation or tensile strength) of the resin composition to be predicted.

[0062] (Operation and Effects of the Embodiment) As described above, in the method for predicting the durability properties of a resin composition according to this embodiment, a regression model 7 is created by machine learning using at least the amounts of base polymer and filler as explanatory variables and the elongation retention rate or tensile strength retention rate as the objective variable, and the elongation retention rate or tensile strength retention rate of the resin composition to be predicted is predicted using the regression model 7.

[0063] By directly predicting the elongation retention rate or tensile strength retention rate using these as the dependent variable, it becomes possible to predict these rates with high accuracy. Furthermore, by adding composite characteristic values ​​or initial physical properties (initial elongation or tensile strength) as explanatory variables, even greater accuracy can be achieved.

[0064] (Second Embodiment) In the first embodiment described above, a direct method was employed in which machine learning was used to directly predict the elongation retention rate or tensile strength retention rate. In contrast, the second embodiment employs an indirect method in which machine learning is used to predict the elongation and tensile strength after the durability test, and based on the obtained elongation and tensile strength after the durability test, the above equations (1) and (2) are used to predict the elongation retention rate or tensile strength retention rate (i.e., indirect prediction). In the second embodiment, although the prediction accuracy is lower compared to the first embodiment, the prediction accuracy is improved by using composite characteristic values ​​as explanatory variables.

[0065] Figure 10 is a schematic diagram of the durability property prediction device 1a (hereinafter simply referred to as durability property prediction device 1a) of a resin composition according to the second embodiment. Since the durability property prediction device 1a has basically the same configuration as the durability property prediction device 1 in Figure 1, a description of the similar configuration will be omitted.

[0066] The durability property prediction device 1a replaces the regression model creation processing unit 22 of the durability property prediction device 1 in Figure 1 with a regression model creation processing unit 22a for post-test property prediction and a regression model creation processing unit 22b for initial property prediction. Furthermore, the durability property prediction processing unit 23 has a post-test property prediction processing unit 23a and an initial property prediction processing unit 23b.

[0067] (Processing unit 22a for creating regression models for predicting physical properties after testing) The regression model creation processing unit 22a for predicting post-test physical properties performs a regression model creation process for predicting post-test physical properties by performing machine learning with at least the respective amounts of base polymer and filler and (a) composite characteristic values ​​including volume fraction parameters as explanatory variables, and elongation or tensile strength after durability testing (hereinafter referred to as post-test physical properties) as the objective variable.

[0068] As shown in Figure 11(a), the regression model creation processing unit 22a for predicting post-test physical properties receives the training data 6 as input. The blending amount, composite characteristic value, and irradiation dose are used as explanatory variables, and the post-test physical property (elongation or tensile strength after the durability test) is used as the objective variable. Based on the input training data 6, the regression model creation processing unit 22a creates a regression model 7a for predicting post-test physical properties that represents the correlation between the explanatory variables (blending amount, composite characteristic value, and irradiation dose) and the objective variable (post-test physical property, elongation or tensile strength after the durability test). The created regression model 7a for predicting post-test physical properties is stored in the storage unit 3.

[0069] (Post-test physical property prediction processing unit 23a) The post-test property prediction processing unit 23a performs post-test property prediction processing to predict the post-test properties (elongation or tensile strength after durability testing) of the resin composition to be predicted, using a regression model 7a for post-test property prediction.

[0070] As shown in Figure 11(b), in the post-test property prediction process, the post-test property prediction processing unit 23a receives the regression model 7a for post-test property prediction and the source data 8 for prediction. The post-test property prediction processing unit 23a uses the regression model 7a for post-test property prediction to determine the post-test property corresponding to the source data 8, and stores the resulting post-test property data 16a, which is the predicted post-test property data 16a, as prediction data 9 in the storage unit 3.

[0071] (Processing unit 22b for creating regression models for initial material property prediction) The regression model creation processing unit 22b for initial physical property prediction performs a regression model creation process for initial physical property prediction, using at least the respective amounts of base polymer and filler and (a) composite characteristic values ​​including volume fraction parameters as explanatory variables, and initial physical properties (initial elongation or tensile strength) as the objective variable, and performing machine learning to create a regression model 7b for initial physical property prediction.

[0072] As shown in Figure 12(a), the regression model creation processing unit 22b for initial physical property prediction receives the training data 6 as input, with the blending amount, composite characteristic value, and irradiation dose used as explanatory variables, and the initial physical property (initial elongation or tensile strength) used as the objective variable. Based on the input training data 6, the regression model creation processing unit 22b creates a regression model 7b for initial physical property prediction that represents the correlation between the explanatory variables (blending amount, composite characteristic value, and irradiation dose) and the objective variable (initial physical property, initial elongation or tensile strength). The created regression model 7b for initial physical property prediction is stored in the storage unit 3.

[0073] (Initial physical property prediction processing unit 23b) The initial physical property prediction processing unit 23b performs initial physical property prediction processing to predict the initial physical properties (initial elongation or tensile strength) of the resin composition to be predicted, using the regression model 7b for initial physical property prediction.

[0074] As shown in Figure 12(b), in the initial physical property prediction process, the initial physical property prediction processing unit 23b receives the regression model 7b for initial physical property prediction and the source data 8 as input. The initial physical property prediction processing unit 23b uses the regression model 7b for initial physical property prediction to determine the initial physical properties corresponding to the source data 8, and stores the resulting initial physical property data 14b, which is the predicted initial physical property data 14b, as predicted data 9 in the storage unit 3.

[0075] (Durability Prediction Processing Unit 23) In the durability property prediction device 1a, the durability property prediction processing unit 23 uses the above equations (1) and (2) to determine the elongation retention rate or tensile strength retention rate based on the predicted post-test property data 16a and the predicted initial property data 14b, which are stored in the storage unit 3 as prediction data 9. The obtained predicted durability property data 15a, which is the elongation retention rate or tensile strength retention rate data, is stored in the storage unit 3 as prediction data 9 (see Figure 10).

[0076] In this embodiment, machine learning-based predicted values ​​were used as initial physical properties, but measured values ​​may also be used as initial physical properties. This will allow for further improvement of prediction accuracy. In this case, the regression model creation processing unit 22b and the initial physical property prediction processing unit 23b can be omitted.

[0077] (Investigation of the effects of using composite property values) Using the same data as used in the study in Figure 7, the indirect method was evaluated using the durability property prediction device 1a, and the coefficient of determination R 2 The mean absolute error was calculated and designated as Example 5. All of (a) to (f) above were used as the composite characteristic values, and the evaluation method was the same as in Figure 7. The comparative example, in which only the blending amount was used as the explanatory variable, was also evaluated in the same manner. The results are shown in Figure 13. In addition, Examples 1 and 2 (the same as Examples 1 and 2 in Figure 7) using the direct method are also shown in Figure 13. Although not shown in the illustration, process data 13 (irradiation dose) was also included as an explanatory variable in all Examples 1, 2, 5 and the comparative example.

[0078] As shown in Figure 13, in Example 5 of this embodiment, the prediction accuracy is lower compared to Examples 1 and 2, which use direct methods, but it is higher than the comparative example that uses only the blending amount. Thus, even when using an indirect method, it is possible to improve prediction accuracy by including the composite characteristic value as an explanatory variable.

[0079] (Summary of the embodiments) Next, the technical concept understood from the embodiments described above will be described using the reference numerals and other symbols from the embodiments. However, the reference numerals and other symbols in the following description are not limited to the components in the claims that are specifically shown in the embodiments.

[0080] [1] A method for predicting the durable physical properties of a resin composition formed using a material comprising a base polymer and a filler, wherein the durable physical properties are the elongation residual rate or tensile strength residual rate shown in the following formula, Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) A method for predicting the durability properties of a resin composition, comprising creating a regression model (7) by performing machine learning with at least the amounts of the base polymer and the filler used as explanatory variables and the elongation retention rate or tensile strength retention rate as the objective variable, and using the regression model (7) to predict the elongation retention rate or tensile strength retention rate of the resin composition to be predicted.

[0081] [2] The explanatory variables include at least the following (a) (a) A volume fraction parameter including information on the ratio of the volume of the base polymer to the volume of the filler. A method for predicting the durability of a resin composition according to [1], using synthetic property values ​​including the following.

[0082] [3] A method for predicting the durability of a resin composition according to [1], further comprising the initial elongation or tensile strength of the resin composition as the explanatory variable.

[0083] [4] A method for predicting the durable properties of a resin composition formed using a material comprising a base polymer and a filler, wherein the durable properties are the elongation percentage or tensile strength percentage shown in the following formula, Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) At least the respective amounts of the base polymer and the filler and the following (a) (a) A volume fraction parameter including information on the ratio of the volume of the base polymer to the volume of the filler. A method for predicting the durability properties of a resin composition, comprising: creating a regression model (7a) for predicting post-test physical properties by performing machine learning with synthetic characteristic values ​​including as explanatory variables and elongation or tensile strength after the durability test as the dependent variable; predicting the elongation or tensile strength of the target resin composition after the durability test using the regression model (7a); and predicting the elongation retention rate or tensile strength retention rate of the target resin composition based on the predicted elongation or tensile strength after the durability test.

[0084] [5] A regression model for predicting initial physical properties (7b) is created by performing machine learning with at least the respective amounts of the base polymer and the filler and the synthetic property values ​​as explanatory variables and the initial elongation or tensile strength as the dependent variable. A method for predicting the durability properties of a resin composition according to [4], wherein the initial elongation or tensile strength of the resin composition to be predicted is predicted using the regression model (7b) for predicting initial physical properties, and the remaining elongation rate or tensile strength rate of the resin composition to be predicted is predicted using the predicted initial elongation or tensile strength.

[0085] [6] The composite characteristic values ​​are, in addition to (a) above, the following (b) to (f) (b) Filler surface area including information on the BET specific surface area of ​​the filler, (c) Amount of maleic anhydride modification, which is the proportion of maleic anhydride contained in the resin composition. (d) The amount of crystalline polymer, which is the ratio of the amount of crystalline polymer to the amount of base polymer. (e) Copolymerization amount, which is the proportion of copolymer components with ethylene contained in the base polymer. (f) Filler identification information, which is information on the type of flame retardant contained in the filler and the surface treatment method of the flame retardant. A method for predicting the durability of a resin composition according to [2] or [4], comprising one or more of the following:

[0086] [7] A method for predicting the durability of a resin composition according to [1] or [4], further comprising process data (13) which are the manufacturing conditions of the resin composition, as explanatory variables.

[0087] [8] A method for predicting the durability of a resin composition according to [7], wherein the resin composition is a crosslinked resin composition that is crosslinked by electron beam irradiation, and the process data (13) is the amount of electron beam irradiation during crosslinking.

[0088] [9] A device for predicting the durable properties of a resin composition formed using a material comprising a base polymer and a filler, wherein the durable properties are the elongation percentage or tensile strength percentage shown in the following formula, Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) A resin composition durability property prediction device (1) comprises: a regression model creation processing unit (22) that creates a regression model (7) by performing machine learning with at least the amounts of the base polymer and the filler as explanatory variables and the elongation retention rate or the tensile strength retention rate as the objective variable; and a durability property prediction processing unit (23) that uses the regression model (7) to predict the elongation retention rate or tensile strength retention rate of the resin composition to be predicted.

[0089]

[10] A device for predicting the durable physical properties of a resin composition formed using a material comprising a base polymer and a filler, wherein the durable physical properties are the elongation remaining rate or tensile strength remaining rate shown in the following formula, Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) At least the respective amounts of the base polymer and the filler and the following (a) (a) A volume fraction parameter including information on the ratio of the volume of the base polymer to the volume of the filler. A resin composition durability property prediction device (1a) comprising: a regression model creation processing unit (22a) that creates a regression model (7a) for predicting post-test physical properties by performing machine learning with synthetic characteristic values ​​including as explanatory variables and elongation or tensile strength after the durability test as the objective variable; a post-test physical property prediction processing unit (23a) that uses the post-test physical property prediction regression model (7a) to predict the elongation or tensile strength of the target resin composition after the durability test; and a durability property prediction processing unit (23) that predicts the elongation retention rate or tensile strength retention rate of the target resin composition based on the elongation or tensile strength after the durability test predicted by the post-test physical property prediction processing unit (23a).

[0090] (Note) Although embodiments of the present invention have been described above, the embodiments described above do not limit the invention as defined in the claims. Furthermore, it should be noted that not all combinations of features described in the embodiments are necessarily essential for solving the problem of the invention. Moreover, the present invention can be implemented with appropriate modifications without departing from its spirit. [Explanation of Symbols]

[0091] 1,1a... Apparatus for predicting the durability properties of resin compositions (durability property prediction apparatus) 2…Control Unit 3...Storage section 6…Training data 7…Regression Model 7a…Regression model for predicting physical properties after testing 7b…Regression model for predicting initial physical properties 8…Prediction source data 9…Predictive data 11…Ingredient amount data 12…Composite characteristic value data 13…Process data 14…Initial physical property data 15…Durability and physical properties data 21...Data acquisition processing unit 22…Regression Model Creation Processing Unit 22a...Processing unit for creating regression models for predicting physical properties after testing 22b…Processing unit for creating regression models for initial physical property prediction 23…Durability and Physical Properties Prediction Processing Unit 23a...Post-test physical property prediction processing unit 23b...Initial physical property prediction processing unit

Claims

1. A method for predicting the durable properties of a resin composition formed using a material containing a base polymer and a filler, which are the physical properties after a predetermined durability test, The aforementioned durable material is the elongation rate or tensile strength rate shown in the following formula: Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) A regression model is created by performing machine learning with at least the amounts of the base polymer and the filler, and the combined characteristic values ​​related to elongation or tensile strength of the entire resin composition as explanatory variables, and the elongation retention rate or tensile strength retention rate as the dependent variable. The aforementioned composite characteristic values ​​are as follows (a) to (e): (a) A volume fraction parameter including information on the ratio of the volume of the base polymer to the volume of the filler. (b) Filler surface area including information on the BET specific surface area of ​​the filler (c) Amount of maleic anhydride modification, which is the proportion of maleic anhydride contained in the resin composition. (d) The amount of crystalline polymer, which is the ratio of the amount of crystalline polymer to the amount of base polymer. (e) Copolymerization amount, which is the proportion of copolymer components with ethylene contained in the base polymer. Includes, Using the regression model described above, predict the elongation retention rate or tensile strength retention rate of the resin composition to be predicted. A method for predicting the durability properties of a resin composition.

2. Furthermore, the initial elongation or tensile strength of the resin composition is used as the explanatory variable. A method for predicting the durability properties of a resin composition according to claim 1.

3. A method for predicting the durable properties of a resin composition formed using a material containing a base polymer and a filler, which are the physical properties after a predetermined durability test, The aforementioned durable material is the elongation rate or tensile strength rate shown in the following formula: Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) A regression model for predicting post-test physical properties is created by performing machine learning with at least the amounts of the base polymer and the filler, and the combined characteristic value obtained by combining the characteristic values ​​related to elongation or tensile strength for the entire resin composition as explanatory variables, and the elongation or tensile strength after the durability test as the dependent variable. The aforementioned composite characteristic values ​​are as follows (a) to (e): (a) A volume fraction parameter including information on the ratio of the volume of the base polymer to the volume of the filler. (b) Filler surface area including information on the BET specific surface area of ​​the filler (c) Amount of maleic anhydride modification, which is the proportion of maleic anhydride contained in the resin composition. (d) The amount of crystalline polymer, which is the ratio of the amount of crystalline polymer to the amount of base polymer. (e) Copolymerization amount, which is the proportion of copolymer components with ethylene contained in the base polymer. Includes, Using the regression model for predicting post-test physical properties, the elongation or tensile strength of the resin composition to be predicted after the durability test is predicted. Based on the predicted elongation or tensile strength after the durability test, the remaining elongation or tensile strength of the resin composition to be predicted is predicted. A method for predicting the durability properties of a resin composition.

4. A regression model for predicting initial physical properties is created by performing machine learning with at least the respective amounts of the base polymer and the filler and the synthetic characteristic value as explanatory variables, and the initial elongation or tensile strength as the dependent variable. Using the regression model for predicting initial physical properties, the initial elongation or tensile strength of the resin composition to be predicted is predicted, and using the predicted initial elongation or tensile strength, the remaining elongation rate or tensile strength rate of the resin composition to be predicted is predicted. A method for predicting the durability properties of a resin composition according to claim 3.

5. The aforementioned composite characteristic value is (f) below. (f) Filler identification information, which is information on the type of flame retardant contained in the filler and the surface treatment method of the flame retardant. including, A method for predicting the durability properties of a resin composition according to claim 1 or 3.

6. Furthermore, process data, which is the manufacturing condition of the resin composition, is used as the explanatory variable. A method for predicting the durability properties of a resin composition according to claim 1 or 3.

7. The aforementioned resin composition is a crosslinked resin composition that is crosslinked by electron beam irradiation. The aforementioned process data is the amount of electron beam irradiation during crosslinking. A method for predicting the durability properties of a resin composition according to claim 6.

8. An apparatus for predicting the durable properties of a resin composition formed using a material containing a base polymer and a filler, which are the physical properties after a predetermined durability test. The aforementioned durable material is the elongation rate or tensile strength rate shown in the following formula: Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) A regression model creation processing unit creates a regression model by performing machine learning with at least the amounts of the base polymer and the filler, and the characteristic values ​​related to elongation or tensile strength, as explanatory variables, and the elongation retention rate or tensile strength retention rate as the objective variable. The system includes a durability property prediction processing unit that uses the regression model described above to predict the elongation retention rate or tensile strength retention rate of the resin composition to be predicted, The aforementioned composite characteristic values ​​are as follows (a) to (e): (a) A volume fraction parameter including information on the ratio of the volume of the base polymer to the volume of the filler. (b) Filler surface area including information on the BET specific surface area of ​​the filler (c) Amount of maleic anhydride modification, which is the proportion of maleic anhydride contained in the resin composition. (d) The amount of crystalline polymer, which is the ratio of the amount of crystalline polymer to the amount of base polymer. (e) Copolymerization amount, which is the proportion of copolymer components with ethylene contained in the base polymer. including, A device for predicting the durability and physical properties of resin compositions.

9. An apparatus for predicting the durable properties of a resin composition formed using a material containing a base polymer and a filler, which are the physical properties after a predetermined durability test. The aforementioned durable material is the elongation rate or tensile strength rate shown in the following formula: Elongation rate = 100 × (Elongation after the durability test / Initial elongation) Tensile strength remaining percentage = 100 × (Tensile strength after the durability test / Initial tensile strength) A processing unit for creating a regression model for predicting post-test physical properties uses at least the respective amounts of the base polymer and the filler, and the characteristic values ​​related to elongation or tensile strength, which are combined for the entire resin composition, as explanatory variables, and performs machine learning to create a regression model for predicting post-test physical properties, with the elongation or tensile strength after the durability test as the objective variable. A post-test physical property prediction processing unit that uses the aforementioned regression model for predicting post-test physical properties to predict the elongation or tensile strength of the resin composition to be predicted after a durability test, The system includes a durability property prediction processing unit that predicts the remaining elongation or tensile strength of the resin composition to be predicted based on the elongation or tensile strength after the durability test predicted by the post-test property prediction processing unit, The aforementioned composite characteristic values ​​are as follows (a) to (e): (a) A volume fraction parameter including information on the ratio of the volume of the base polymer to the volume of the filler. (b) Filler surface area including information on the BET specific surface area of ​​the filler (c) Amount of maleic anhydride modification, which is the proportion of maleic anhydride contained in the resin composition. (d) The amount of crystalline polymer, which is the ratio of the amount of crystalline polymer to the amount of base polymer. (e) Copolymerization amount, which is the proportion of copolymer components with ethylene contained in the base polymer. including, A device for predicting the durability and physical properties of resin compositions.

Images