Model generation device, prediction device, model generation method, prediction method, and resin composition manufacturing system

Abstract

To find a condition for satisfying required characteristics of a resin composition more efficiently than before.SOLUTION: In a model generation device (100), a first machine learning unit (21) generates a first prediction model (MODEL1) for predicting unknown resin composition characteristic data from at least one of (i) optional inorganic filler characteristic data, (ii) optional resin characteristic data, (iii) optional inorganic filler blending data, and (iv) optional resin blending data, based on first input data (110) and second input data (120) paired with the first input data (110). A second machine learning unit (22) generates a second prediction model (MODEL2) for predicting at least one of (i) predicted inorganic filler characteristic data, (ii) predicted resin characteristic data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data, which satisfy arbitrary resin composition characteristic data, based on the first prediction model (MODEL1).SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] One aspect of the present invention relates to a model generation device that generates a prediction model for predicting the required property satisfaction conditions (described later) for a resin composition.

Background Art

[0002] In order to obtain a resin composition having desired required properties (performance), it is necessary to find the conditions of the resin composition that satisfy the required properties (e.g., the composition of the resin composition). However, conventionally, in order to find such conditions, a large number of trial-and-error composition studies based on experiments have been required. For example, Patent Document 1 shows that it was necessary to conduct a large number of experiments to find a suitable composition of an epoxy resin composition that satisfies the required property of "being able to suppress bleeding".

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] An object of one aspect of the present invention is to more efficiently find one or more of (i) the properties of the inorganic filler, (ii) the properties of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, that satisfy the required properties of a resin composition containing one or more inorganic fillers and one or more resins (hereinafter collectively referred to as "required property satisfaction conditions") than in the prior art.

Means for Solving the Problems

[0005] In order to solve the above problems, a model generation device according to one aspect of the present invention A model generation device that generates a prediction model for predicting one or more of (i) the characteristics of the inorganic filler, (ii) the characteristics of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, which satisfy the required characteristics of a resin composition containing one or more inorganic fillers and one or more resins. A first input data acquisition unit that acquires first input data including at least one of (i) inorganic filler characteristic input data indicating the characteristics of the inorganic filler, (ii) resin characteristic input data indicating the characteristics of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the inorganic filler in the resin composition in which a plurality of types of the inorganic filler are blended with the resin, and (iv) resin blending input data regarding the blending ratio of the resin. A second input data acquisition unit that acquires resin composition characteristic input data indicating the characteristics of the resin composition as second input data paired with the first input data. A first machine learning unit that generates a first prediction model for predicting unknown resin composition characteristic data from one or more of (i) arbitrary inorganic filler characteristic data, (ii) arbitrary resin characteristic data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data, based on the first input data and the second input data. A second machine learning unit that generates a second prediction model for predicting one or more of (i) predicted inorganic filler characteristic data, (ii) predicted resin characteristic data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data, which satisfy arbitrary resin composition characteristic data, based on the first prediction model.

[0006] Further, a prediction device according to an aspect of the present invention A prediction device that predicts one or more of (i) the characteristics of the inorganic filler, (ii) the characteristics of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, which satisfy the required characteristics of a resin composition containing one or more inorganic fillers and one or more resins. (i) Inorganic filler property input data indicating the properties of the inorganic filler, (ii) resin property input data indicating the properties of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the plurality of types of inorganic fillers in the resin composition obtained by blending the inorganic fillers into the resin, and (iv) resin blending input data regarding the blending ratio of the resin, at least one of which is included in the first input data, is acquired in advance. As the second input data paired with the first input data, resin composition property input data indicating the properties of the resin composition is acquired in advance. A first prediction model for predicting unknown resin composition property data from one or more of (i) arbitrary inorganic filler property data, (ii) arbitrary resin property data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data is generated in advance based on the first input data and the second input data. A second prediction model for predicting one or more of (i) predicted inorganic filler property data, (ii) predicted resin property data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data that satisfy arbitrary resin composition property data is generated in advance based on the first prediction model. The prediction device A third input data acquisition unit that acquires, as third input data, resin composition required property data indicating the required properties of the resin composition. By inputting the resin composition required property data into the second prediction model, a recommended data including at least one of (i) recommended inorganic filler property data indicating the properties of the inorganic filler that satisfies the resin composition required property data, (ii) recommended resin property data indicating the properties of the resin that satisfies the resin composition required property data, (iii) recommended inorganic filler blending data indicating the blending of the inorganic filler that satisfies the resin composition required property data, and (iv) recommended resin blending data indicating the blending of the resin that satisfies the resin composition required property data is derived by a recommended data derivation unit.

[0007] Further, a model generation method according to one aspect of the present invention A model generation method for generating a prediction model for predicting one or more of (i) the properties of the inorganic filler, (ii) the properties of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, which satisfies the required properties of a resin composition containing one or more inorganic fillers and one or more resins. A first input data acquisition step of acquiring first input data which is input data including at least one of (i) inorganic filler property input data indicating the properties of the inorganic filler, (ii) resin property input data indicating the properties of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the inorganic filler in the resin composition in which a plurality of types of the inorganic filler are blended into the resin, and (iv) resin blending input data regarding the blending ratio of the resin. A second input data acquisition step of acquiring resin composition property input data indicating the properties of the resin composition as second input data paired with the first input data. A first machine learning step of generating a first prediction model for predicting unknown resin composition property data from one or more of (i) arbitrary inorganic filler property data, (ii) arbitrary resin property data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data, based on the first input data and the second input data. A second machine learning step of generating a second prediction model for predicting one or more of (i) predicted inorganic filler property data, (ii) predicted resin property data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data, which satisfy arbitrary resin composition property data, based on the first prediction model.

[0008] Further, a prediction method according to one aspect of the present invention is A prediction method for predicting one or more of (i) the properties of the inorganic filler, (ii) the properties of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, which satisfies the required properties of a resin composition containing one or more inorganic fillers and one or more resins. (i) Inorganic filler property input data indicating the properties of the inorganic filler, (ii) resin property input data indicating the properties of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the plurality of types of inorganic fillers in the resin composition obtained by blending the inorganic fillers into the resin, and (iv) resin blending input data regarding the blending ratio of the resin, and the first input data, which is input data including at least one of them, has been acquired in advance. As the second input data corresponding to the first input data, resin composition property input data indicating the properties of the resin composition has been acquired in advance. A first prediction model for predicting unknown resin composition property data from one or more of (i) arbitrary inorganic filler property data, (ii) arbitrary resin property data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data has been generated in advance based on the first input data and the second input data. A second prediction model for predicting one or more of (i) predicted inorganic filler property data, (ii) predicted resin property data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data that satisfy arbitrary resin composition property data has been generated in advance based on the first prediction model. The prediction method includes: A third input data acquisition step of acquiring, as third input data, resin composition required property data indicating the required properties of the resin composition; A recommended data derivation step of inputting the resin composition required property data into the second prediction model to derive recommended data including at least one of (i) recommended inorganic filler property data indicating the properties of the inorganic filler that satisfies the resin composition required property data, (ii) recommended resin property data indicating the properties of the resin that satisfies the resin composition required property data, (iii) recommended inorganic filler blending data indicating the blending of the inorganic filler that satisfies the resin composition required property data, and (iv) recommended resin blending data indicating the blending of the resin that satisfies the resin composition required property data.

[0009] Moreover, a resin composition manufacturing system according to one aspect of the present invention A model generation device that generates a prediction model for predicting one or more of (i) the properties of the inorganic filler, (ii) the properties of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, which satisfy the required properties of a resin composition containing one or more inorganic fillers and one or more resins. A prediction device that predicts one or more of (i) the properties of the inorganic filler, (ii) the properties of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, which satisfy the required properties of the resin composition, using the prediction model generated by the model generation device. The model generation device (i) A first input data acquisition unit that acquires first input data, which is input data including at least one of (i) inorganic filler property input data indicating the properties of the inorganic filler, (ii) resin property input data indicating the properties of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the inorganic filler in the resin composition in which a plurality of types of the inorganic filler are blended into the resin, and (iv) resin blending input data regarding the blending ratio of the resin. A second input data acquisition unit that acquires resin composition property input data indicating the properties of the resin composition as second input data paired with the first input data. (i) A first machine learning unit that generates a first prediction model for predicting unknown resin composition property data from one or more of (i) arbitrary inorganic filler property data, (ii) arbitrary resin property data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data, based on the first input data and the second input data. A second machine learning unit that generates a second prediction model for predicting one or more of (i) predicted inorganic filler property data, (ii) predicted resin property data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data, which satisfy arbitrary resin composition property data, based on the first prediction model. The prediction device A third input data acquisition unit that acquires, as third input data, resin composition required property data indicating the required properties of the resin composition; By inputting the resin composition required property data into the second prediction model, (i) recommended inorganic filler property data indicating the properties of an inorganic filler that satisfies the resin composition required property data, (ii) recommended resin property data indicating the properties of a resin that satisfies the resin composition required property data, (iii) recommended inorganic filler blending data indicating the blending of an inorganic filler that satisfies the resin composition required property data, and (iv) recommended resin blending data indicating the blending of a resin that satisfies the resin composition required property data, a recommended data derivation unit that derives recommended data including at least one of the above.

Advantages of the Invention

[0010] According to one aspect of the present invention, it becomes possible to find the required property satisfaction conditions for a resin composition more efficiently than in the past.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Mode for Carrying Out the Invention

[0012] 〔Embodiment 1〕 The resin composition manufacturing system 1 of Embodiment 1 will be described below. For the sake of convenience of explanation, components having the same functions as the components (components) described in Embodiment 1 are given the same reference numerals, and the description thereof will not be repeated. For the sake of simplification, descriptions of matters similar to known technologies are also omitted as appropriate. It should be noted that each configuration and each numerical value described in this specification are merely examples unless otherwise specified. In this specification, the description "X~Y" regarding two numbers X and Y means "X or more and Y or less" unless otherwise specified.

[0013] (Outline of the resin composition manufacturing system 1) FIG. 1 is a block diagram showing the configuration of the main part of the resin composition manufacturing system 1. The resin composition manufacturing system 1 includes a model generation device 100 and a prediction device 300. As will be described later, the model generation device 100 generates a prediction model for predicting the required characteristic satisfaction conditions for a resin composition containing one or more inorganic fillers and one or more resins. Specifically, the model generation device 100 generates a second prediction model (MODEL2) described later as the prediction model. Then, the prediction device 300 predicts the required characteristic satisfaction conditions using MODEL2.

[0014] In the following description, one or more inorganic fillers are generically referred to as inorganic filler A, and one or more resins are generically referred to as resin B. And a resin composition containing inorganic filler A and resin B is referred to as resin composition C.

[0015] As main examples of the inorganic filler A, ceramics (e.g., silica, alumina, aluminum nitride, silicon nitride, and boron nitride) can be cited. The inorganic filler is also referred to as an inorganic filler. By changing the addition amount of the inorganic filler A to the resin composition C, the properties of the resin composition C can be changed. For example, by appropriately setting the addition amount of the inorganic filler A, the elastic modulus and the linear expansion coefficient of the resin composition C can be improved. In Embodiment 1, the case where the inorganic filler A is silica is mainly exemplified.

[0016] As main examples of the resin B, any organic polymer compound can be cited. The resin composition C contains the inorganic filler A and the resin B, but may further contain other materials (e.g., curing agent, elastomer, ion trap agent, pigment, dye, antifoaming agent, stress reliever, pH adjuster, accelerator, surfactant, and coupling agent). As main examples of the resin composition C, adhesives and sealants can be cited.

[0017] Note that it should be noted that the "required property satisfaction condition" in this specification is not limited to the condition that "completely satisfies the required properties of the resin composition C". The "required property satisfaction condition" in this specification includes, for example, "the required property closest to the complete required properties of the resin composition C" and "the required property somewhat close to the complete required properties of the resin composition C". Therefore, the resin composition manufacturing system 1 only needs to be able to predict, as the required property satisfaction condition, the condition that generally satisfies the complete required properties of the resin Composition C. For this reason, for example, as described later, the resin composition manufacturing system 1 may predict the required property satisfaction condition based on a probabilistic mathematical model.

[0018] (Model generation device 100) The processing of the resin composition manufacturing system 1 is roughly classified into a learning phase (processing in the model generation device 100) and a prediction phase (processing in the prediction device 300). The prediction phase is also referred to as an inference phase. First, the learning phase will be described. The model generation device 100 includes a first input data acquisition unit 11, a second input data acquisition unit 12, a first machine learning unit 21, and a second machine learning unit 22.

[0019] The first input data acquisition unit 11 acquires the first input data 110. The first input data 110 is · Inorganic filler property input data 111, · Resin property input data 112, · Inorganic filler blending input data 113, and · Resin blending input data 114, input data including one of them.

[0020] The inorganic filler property input data 111 indicates the properties of the inorganic filler A. The resin property input data 112 indicates the properties of the resin B. The inorganic filler blending input data 113 is data regarding the blending ratio of the plurality of types of inorganic filler A in the resin composition C obtained by blending the inorganic filler A into the resin B. As an example, the inorganic filler blending input data 113 is data indicating the blending ratio of the inorganic filler A. The resin blending input data 114 is data regarding the blending ratio of the resin B in the resin composition C. As an example, the resin blending input data 114 is data indicating the blending ratio of the resin B.

[0021] The first input data acquisition unit 11 may be an arbitrary data acquisition interface. As an example, the first input data acquisition unit 11 may be an input unit that receives a user's input operation. In this case, the first input data acquisition unit 11 acquires the first input data 110 input by the user. As another example, the first input data acquisition unit 11 may acquire the first input data 110 previously stored in a storage unit (not shown) in the model generation device 100. As yet another example, the first input data acquisition unit 11 may communicate with an external device (not shown) of the model generation device 100 and acquire the first input data 110 from the external device. Examples of the external device may include a storage server and a measuring device. These explanations for the first input data acquisition unit 11 also apply equally to the second input data acquisition unit 12 described below and the third input data acquisition unit 33 described later.

[0022] The second input data acquisition unit 12 acquires the second input data 120. Specifically, the second input data acquisition unit 12 acquires resin composition property input data as the second input data 120. The resin composition property input data indicates the properties of the resin composition C. From this, it can be said that the second input data 120 is data that pairs with the first input data 110.

[0023] The first machine learning unit 21 acquires the first input data 110 from the first input data acquisition unit 11 and the second input data 120 from the second input data acquisition unit 12, respectively. The first machine learning unit 21 generates a first prediction model (hereinafter referred to as MODEL1) based on the first input data 110 and the second input data 120.

[0024] FIG. 2 is a diagram for explaining the outline of MODEL1. As shown in FIG. 2, MODEL1 is · Any inorganic filler property data 1110, · Any resin property data 1120, · Any inorganic filler blending data 1130, and · Any resin blending data 1140, is a model (mathematical model) for predicting unknown resin composition property data 1200 from one or more of them. In this specification, a data set including one or more of any inorganic filler property data 1110, any resin property data 1120, any inorganic filler blending data 1130, and any resin blending data 1140 is referred to as an arbitrary input data set 1100.

[0025] In the example of FIG. 2, the arbitrary input data set 1100 is an example of an explanatory variable (X). And the unknown resin composition property data 1200 is an example of an objective variable (y). Note that an explanatory variable is also referred to as an independent variable. In contrast, an objective variable is also referred to as a dependent variable or an explained variable. In the example of FIG. 2, MODEL1 can be represented as a function f showing the relationship of y = f(X). Thus, MODEL1 is a model for solving the forward problem (a model for deriving y from X).

[0026] In addition, in this specification, it is assumed that the types of data included in any input dataset 1100 match the types of teacher data used to generate MODEL1. That is, it is assumed that the data structure of any input dataset 1100 matches the data structure of the teacher data used to generate MODEL1. Therefore, for example, the data structure of any input dataset 1100 matches the data structure of the first input data 110.

[0027] The second machine learning unit 22 acquires MODEL1 from the first machine learning unit 21. The second machine learning unit 22 generates a second prediction model (hereinafter referred to as MODEL2) based on MODEL1.

[0028] FIG. 3 is a diagram for explaining the outline of MODEL2. As shown in FIG. 3, MODEL2 is a model for predicting one or more of: · Predicted inorganic filler property data 2310, · Predicted resin property data 2320, · Predicted inorganic filler blending data 2330, and · Predicted resin blending data 2340, that satisfy any resin composition property data 2200. Note that the description "satisfy any resin composition property data 2200" in this specification means "satisfy the required properties of the resin composition C indicated by any resin composition property data 2200". Also, in this specification, a dataset including one or more of the predicted inorganic filler property data 2310, the predicted resin property data 2320, the predicted inorganic filler blending data 2330, and the predicted resin blending data 2340 is referred to as a prediction dataset 2300.

[0029] In the example of FIG. 3, any resin composition property data 2200 is an example of the target variable (y). And the prediction dataset 2300 is an example of the explanatory variable (X). In the example of FIG. 3, MODEL2 can be represented as a function g showing the relationship of X = g(y). Note that g ≈ f -1That is, the function g is an approximate inverse function of the function f. Thus, MODEL2 is a model for solving an inverse problem (a model for deriving X from y). As described above, MODEL2 is a model paired with MODEL1.

[0030] In this specification, it is assumed that the type of any resin composition characteristic data 2200 matches the type of the teacher data used for generating MODEL2. That is, it is assumed that the data structure of any resin composition characteristic data 2200 matches the data structure of the teacher data used for generating MODEL2. Therefore, for example, the data structure of any resin composition characteristic data 2200 matches the data structure of the second input data 120.

[0031] In the example of FIG. 1, the first machine learning unit 21 has a first algorithm execution unit 211 and a second algorithm execution unit 212. The first algorithm execution unit 211 acquires the first input data 110 from the first input data acquisition unit 11. The first algorithm execution unit 211 executes a first algorithm for deriving an explanatory variable (X) corresponding to data indicating the characteristics of the resin composition C obtained from the second input data 120. Examples of the first algorithm will be described later.

[0032] The second algorithm execution unit 212 acquires the explanatory variable (X) from the first algorithm execution unit 211 and acquires the second input data 120 from the second input data acquisition unit 12. The second algorithm execution unit 212 executes a second algorithm for generating MODEL1 based on X and the second input data 120.

[0033] The second algorithm may be any algorithm capable of generating MODEL1 based on X and the second input data 120. In other words, the second algorithm may be any algorithm capable of generating a model for solving a forward problem. As an example, the second algorithm is · Gaussian process regression, · Support vector machine, · Linear regression, · Decision tree, · Random forest, · Neural network, and · Gradient boosting tree is at least one of them.

[0034] In the example of FIG. 1, the second machine learning unit 22 has a third algorithm execution unit 223. The third algorithm execution unit 223 executes a third algorithm for generating MODEL2 based on MODEL1.

[0035] The third algorithm may be any algorithm capable of generating MODEL2 based on MODEL1. In other words, the third algorithm may be any algorithm capable of generating a model for solving an inverse problem. As an example, the third algorithm is · Genetic algorithm, · Steepest descent method, · Grid search, and · Bayesian optimization is at least one of them.

[0036] In one aspect of the present invention, the inorganic filler property input data 111 may be data indicating any property of the inorganic filler A. As an example, the inorganic filler property input data 111 is the composition formula, crystallinity, specific gravity, bulk density, particle size distribution, specific surface area, pore volume, zeta potential, electrical conductivity, dielectric constant, dielectric loss tangent, refractive index, specific heat, thermal conductivity, linear expansion coefficient, crushing strength, sphericity, aspect ratio, moisture content, carbon content, nitrogen content, surface functional group type, surface functional group amount, light absorption wavelength, absorbance, M value, and solubility parameter of at least one of them as the property of the inorganic filler A. Note that the description regarding the inorganic filler property input data 111 also applies equally to each data corresponding to the inorganic filler property input data 111.

[0037] In one aspect of the present invention, the resin property input data 112 may be data indicating any property of resin B. As an example, the resin property input data 112 is of resin B, constituent formula, degree of polymerization, molecular weight distribution, stereoregularity, reactive functional group type, amount of reactive functional group, viscosity, melting point, glass transition temperature, crystallinity, elastic modulus, yield stress, tensile strength, fracture toughness, light absorption wavelength, absorbance, specific gravity, refractive index, electrical conductivity, dielectric constant, dielectric loss tangent, specific heat, thermal conductivity, moisture content, and solubility parameter, shows at least one of them as a property of resin B. Note that the description regarding the resin property input data 112 also applies equally to each data corresponding to the resin property input data 112.

[0038] In one aspect of the present invention, the resin composition property input data (second input data 120) may be data indicating any property of the resin composition C. As an example, the resin composition property input data is of the resin composition C, viscosity, fluidity, moldability, adhesiveness, transparency, color tone, strength, water absorption rate, linear expansion coefficient, elastic modulus, yield stress, tensile strength, fracture toughness, electrical conductivity, dielectric constant, dielectric loss tangent, thermal conductivity, and stability, shows at least one of them as a property of the resin composition C. Note that the description regarding the resin composition property input data also applies equally to each data corresponding to the resin composition property input data.

[0039] (Prediction device 300) Subsequently, the prediction phase will be described. The prediction device 300 includes a third input data acquisition unit 33, a recommended data derivation unit 34, and an output unit 35.

[0040] The third input data acquisition unit 33 acquires the third input data 330. Specifically, the third input data acquisition unit 33 acquires the resin composition required property data as the third input data 330. The resin composition required property data indicates the required properties of the resin composition C.

[0041] The recommended data derivation unit 34 acquires the resin composition required property data (third input data 330) from the third input data acquisition unit 33, and acquires MODEL2 from the model generation device 100 (more specifically, the second machine learning unit 22).

[0042] FIG. 4 is a diagram for explaining the outline of the recommended data derivation unit 34. As shown in FIG. 4, the recommended data derivation unit 34 derives the recommended data 340 by inputting the resin composition required property data into MODEL2. The recommended data 340 includes · Recommended inorganic filler property data 341, · Recommended resin property data 342, · Recommended inorganic filler blending data 343, and · Recommended resin blending data 344, including at least one of them.

[0043] The recommended inorganic filler property data 341 indicates the properties of the inorganic filler A that satisfy the resin composition required property data. The recommended resin property data 342 indicates the properties of the resin B that satisfy the resin composition required property data. The recommended inorganic filler blending data 343 indicates the blending of the inorganic filler A that satisfies the resin composition required property data. The recommended resin blending data 344 indicates the blending of the resin B that satisfies the resin composition required property data. Note that the description "satisfies the resin composition required property data" in this specification means "satisfies the required properties of the resin composition C indicated by the resin composition required property data". A specific example of the process of deriving the recommended data 340 by the recommended data derivation unit 34 will be described later.

[0044] As described above, the "required property satisfaction condition" in this specification is not limited to the condition of "completely satisfying the required properties of the resin composition C". Therefore, naturally, the recommended data 340 in this specification is not limited to the data that "completely satisfies the resin composition required property data". The recommended data 340 in this specification also includes the data that "substantially satisfies the resin composition required property data".

[0045] Therefore, the recommended inorganic filler characteristic data 341 in this specification only needs to show the characteristics of the inorganic filler A that generally satisfy the resin composition required characteristic data. Similarly, the recommended resin characteristic data 342 only needs to show the characteristics of the resin B that generally satisfy the resin composition required characteristic data. Also, the recommended inorganic filler blending data 343 only needs to show the blending of the inorganic filler A that generally satisfies the resin composition required characteristic data. Similarly, the recommended resin blending data 344 only needs to show the blending of the resin B that generally satisfies the resin composition required characteristic data. Note that these explanations regarding the recommended inorganic filler characteristic data 341, the recommended resin characteristic data 342, the recommended inorganic filler blending data 343, and the recommended resin blending data 344 also apply equally to the above-described predicted inorganic filler characteristic data 2310, predicted resin characteristic data 2320, predicted inorganic filler blending data 2330, and predicted resin blending data 2340.

[0046] The output unit 35 acquires the recommended data 340 from the recommended data derivation unit 34. Then, the output unit 35 outputs the recommended data 340. The output unit 35 may be any output interface. As an example, the output unit 35 may be a display (display device). In this case, the output unit 35 can visually present the recommended data 340 to the user by displaying the recommended data 340. Thus, the output unit 35 may output the recommended data 340 in a visual manner. As another example, the output unit 35 may transfer the recommended data 340 to a storage unit in the model generation device 100. As yet another example, the output unit 35 may transfer the recommended data 340 to an external device of the model generation device 100.

[0047] In one aspect of the present invention, the resin composition required characteristic data (third input data 330) may be data indicating any required characteristic of the resin composition C. As is clear from the above description of the resin composition characteristic input data, as an example, the resin composition required characteristic data is that of the resin composition C, viscosity, fluidity, moldability, adhesiveness, transparency, color tone, strength, water absorption rate, linear expansion coefficient, elastic modulus, yield stress, tensile strength, fracture toughness, electrical conductivity, dielectric constant, dielectric tangent, thermal conductivity, and stability, At least one of them is shown as a required property of the resin composition C.

[0048] (An example of the process in the resin composition manufacturing system 1) Hereinafter, an example of the process in the resin composition manufacturing system 1 will be described. In the following example, the case where there are five types of inorganic filler A and one type of resin B will be described. In the following example, it is assumed that both the properties of resin B and the blending ratio of resin B are set to constant fixed values (fixed conditions).

[0049] Therefore, in the following example, the properties of resin B and the blending ratio of resin B are not considered as explanatory variables. Therefore, in the following description, the resin property input data 112 and the resin blending input data 114 are not mentioned. From this, hereinafter, the blending ratio of the inorganic filler A is simply abbreviated as the blending ratio. In the following example, as a property of the resin composition C, the viscosity (unit: Pa·s) of the resin composition C is exemplified. Also, the viscosity of the resin composition C is simply abbreviated as viscosity.

[0050] (An example of the first input data 110 and the second input data 120) FIG. 5 is a diagram showing an example of the inorganic filler blending input data 113 and the second input data 120. The inorganic filler blending input data 113 in the example of FIG. 5 is data showing the blending ratios of five types of inorganic fillers in weight% (wt%). In the following description, the five types of inorganic fillers are respectively referred to as inorganic filler 0 to inorganic filler 4. And the blending ratio of the inorganic filler i is referred to as x0i. i is an integer satisfying 0 ≦ i ≦ 4. For example, x01 represents the blending ratio of the inorganic filler 1.

[0051] Here, as is obvious to those skilled in the art, x00 + x01 + x02 + x03 + x04 = 100 The following relationship holds. Therefore, x00 is x00 = 100 - x01 - x02 - x03 - x04 As shown, it is uniquely determined according to the preset x01 to x04. Therefore, in the inorganic filler blending input data 113 in the example of FIG. 5, only x01 to x04 are set in order to reduce the dimensionality of the explanatory variables.

[0052] In the inorganic filler blending input data 113, a plurality of blending ratio patterns (patterns of combinations of x01 to x04) are set. The id in FIG. 5 is the identification number of the blending ratio pattern. For example, id = 1 indicates the first blending ratio pattern (hereinafter also referred to as the first blending ratio pattern). In the first blending ratio pattern in the example of FIG. 5, "x00 = 60, x01 = 10, x02 = 30, x03 = 0, x04 = 0". Hereinafter, for example, id = 1 is appropriately abbreviated as id1.

[0053] The second input data 120 in the example of FIG. 5 indicates the viscosity (y01) of the resin composition C corresponding to each blending ratio pattern. Specifically, in y01 in the second input data 120, the values of the viscosities measured for each blending ratio pattern are recorded. Prior to the computer simulation in this example, the inventors of the present application (hereinafter simply abbreviated as "the inventors") manufactured the resin composition C according to the above-mentioned first blending ratio pattern. When the inventors measured the viscosity of the resin composition C, a value of 455 was obtained as the measured value. Therefore, in the second input data 120 in the example of FIG. 5, a value of y01 = 455 is set for id1.

[0054] In the example of FIG. 5, a dataset showing the correspondence between each blending ratio pattern and y01 is created for each id. In the example of FIG. 5, the dataset corresponding to the j-th id (idj) is referred to as DATASET_idj. As an example, DATASET_id1 is a dataset showing the correspondence between the first blending ratio pattern and y01. Hereinafter, the j-th blending ratio pattern is also referred to as the j-th blending ratio pattern.

[0055] FIG. 6 is a diagram showing an example of the inorganic filler property input data 111. The inorganic filler property input data 111 in the example of FIG. 6 shows the particle size distributions of inorganic fillers 0 to 4.

[0056] (Example of Derivation of Explanatory Variables in the First Algorithm Execution Unit 211) FIG. 7 is a diagram showing an example of the derivation of explanatory variables in the first algorithm execution unit 211. The first algorithm execution unit 211 derives an explanatory variable (X) corresponding to the data indicating the characteristics of the resin composition C obtained from the first input data 110. In this example, the first algorithm execution unit 211 derives an explanatory variable corresponding to the data indicating the viscosity of the resin composition C obtained from the inorganic filler blending input data 113 and the inorganic filler property input data 111. Thus, in this example, the explanatory variable for explaining the viscosity, which is the target variable, is derived as X.

[0057] Specifically, the first algorithm execution unit 211 derives X based on the inorganic filler blending input data 113 and the inorganic filler property input data 111 by executing the first algorithm. In the example of FIG. 7, the first machine learning unit 21 executes weighted average calculation and principal component analysis as the first algorithm.

[0058] In the example of FIG. 7, the first machine learning unit 21 performs weighted averaging on the inorganic filler blending input data 113 based on the inorganic filler property input data 111. Subsequently, the first machine learning unit 21 performs principal component analysis on the weighted-averaged inorganic filler blending input data to derive a particle size distribution-derived vector (a vector having xx01 to xx05 in FIG. 7 as components) as an explanatory variable. Note that the weight value in the weighted average calculation may be set by a known method based on, for example, the inorganic filler property input data 111. The dimension reduced by the principal component analysis may be arbitrarily set. The first machine learning unit 21 calculates a particle size distribution-derived vector for each id. Therefore, for example, as shown in FIG. 7, the explanatory variables derived by the first machine learning unit 21 include the following first particle size distribution-derived vector (the particle size distribution-derived vector corresponding to id1).

[0059] In this specification, the particle size distribution-derived vector corresponding to the j-th blending ratio pattern (in other words, the particle size distribution-derived vector corresponding to idj) is referred to as the j-th particle size distribution-derived vector. The j-th particle size distribution-derived vector is calculated from the particle size distributions of inorganic fillers 0 to 4 in the resin composition C when the j-th blending ratio pattern is applied. Therefore, for example, the first particle size distribution-derived vector is calculated from the particle size distributions of inorganic fillers 0 to 4 in the resin composition C when the above-described first blending ratio pattern is applied.

[0060] (Example of generation of MODEL1 in the second algorithm execution unit 212) FIG. 8 is a diagram showing an example of generation of MODEL1 in the second algorithm execution unit 212. The second algorithm execution unit 212 generates MODEL1 based on (i) the explanatory variable (X) derived by the first algorithm execution unit 211 and (ii) the second input data 120. Specifically, the second algorithm execution unit 212 generates MODEL1 based on X and the second input data 120 by executing the second algorithm.

[0061] In the example of FIG. 8, the second algorithm execution unit 212 executes a neural network as the second algorithm. Specifically, for each id, the second algorithm execution unit 212 acquires the objective variable corresponding to X from the second input data 120. For example, the second algorithm execution unit 212 acquires the viscosity (y01) shown in DATASET_id1 as the correct answer data of the objective variable (y) corresponding to X in id1. Then, the second algorithm execution unit 212 executes a neural network using the correct answer data to derive a function f that satisfies the relationship between X and y for each id. In this way, the second algorithm execution unit 212 generates MODEL1 as a function f indicating the relationship of y = f(X).

[0062] The following exemplifies the case where the second algorithm execution unit 212 performs ensemble learning by the bagging method to generate MODEL1. Therefore, MODEL1 in this example includes a plurality of neural networks (weak learners) having different hyperparameters. Thus, MODEL1 in this example is generated as a strong learner in which a plurality of weak learners are integrated.

[0063] (Example of input and output in MODEL1) FIG. 9 is a diagram showing an example of input and output in MODEL1. In FIG. 9, the above-mentioned first grain size distribution-derived vector (the grain size distribution-derived vector corresponding to id1) is exemplified as X. As shown in FIG. 9, by inputting X into MODEL1, a histogram (y_Hist) showing the distribution of y can be obtained. Specifically, by inputting X into each of the plurality of weak learners in MODEL1 and integrating the plurality of y output from the plurality of weak learners, y_Hist is obtained.

[0064] In this example, prior to the generation of MODEL1 by the above-mentioned second algorithm, preprocessing of the correct data is performed. Specifically, in this example, prior to the generation of MODEL1, logarithmic transformation of the correct data is performed. Therefore, MODEL1 is strictly generated as a model that outputs log(y). Therefore, the horizontal axis in y_Hist in the example of FIG. 9 is log(y). However, in this specification, for the sake of simplicity, it is described as if MODEL1 is a model that outputs y. Further, y_Hist is described by reading it as a histogram showing the distribution of y.

[0065] Then, MODEL1 in this example determines predetermined data based on y_Hist as the final prediction value (the prediction value as a strong learner) and outputs the final prediction value. Specifically, MODEL1 in this example outputs the average value (μ) of y_Hist as the final prediction value (y). μ is also referred to as the expected value.

[0066] According to MODEL1 generated as a reinforcement learner, as shown in FIG. 2 above, by obtaining any input dataset 1100 as an explanatory variable (X), unknown resin composition characteristic data 1200 can be output as an objective variable (y). For example, by inputting the above-described first blending ratio pattern into MODEL1 as X, a predicted value of viscosity corresponding to the first blending ratio pattern can be output as y.

[0067] Note that MODEL1 may output the variance (σ 2 ) of y_Hist together with μ. The variance is an example of an index of the uncertainty of the predicted value of MODEL1. Alternatively, MODEL1 may output the standard deviation (σ) of y_Hist together with μ. The standard deviation is another example of an index of the uncertainty of the predicted value of MODEL1.

[0068] (Calculation example in MODEL2 generated by the third algorithm execution unit 223) FIG. 10 is a diagram showing a calculation example in MODEL2 generated by the third algorithm execution unit 223. The third algorithm execution unit 223 generates MODEL2 based on MODEL1 by executing the third algorithm. In other words, the third algorithm execution unit 223 determines the above-described function g (an approximate inverse function of the function f) based on the function f determined in advance by the second algorithm execution unit 212 (see also FIG. 3 above).

[0069] In the example of FIG. 10, the third algorithm execution unit 223 generates MODEL2 by using a grid search as the third algorithm. Inside the generated MODEL2, the following two-stage (the first stage and the second stage) calculations are executed.

[0070] First, in the first stage, MODEL2 calculates the predicted viscosity values (μ) for each of the plurality of possible compounding ratio patterns (combinations of x01 to x04) using MODEL1. Specifically, MODEL2 inputs the explanatory variables (X) corresponding to each compounding ratio pattern into MODEL1, causing MODEL1 to output μ as the target variable. In this example, MODEL1 further outputs σ in addition to μ for each compounding ratio pattern.

[0071] Subsequently, in the second stage, MODEL2 outputs the compounding ratio pattern (X) for which the probability of obtaining any arbitrary viscosity data input to MODEL2 is maximized based on μ and σ calculated in the first stage (see also FIG. 3 above).

[0072] In the example of FIG. 10, MODEL2 predicts the prediction data set 2300 for which the probability of obtaining any arbitrary resin composition property data 2200 is maximized. In the example of FIG. 10, a case where the arbitrary resin composition property data 2200 is data indicating a viscosity of "μ = 290 to 310" is illustrated.

[0073] The numerical range of μ = 290 to 310 is an example of a numerical range set assuming the case where resin composition C is used as an adhesive. For example, when the viscosity of the adhesive is too high, it is difficult to shape the adhesive into a desired form for use. On the other hand, when the viscosity of the adhesive is too low, the adhesive is likely to drip. Therefore, it is considered that there is a suitable numerical range for the viscosity of the adhesive. The numerical range of μ = 290 to 310 is an example of such a suitable numerical range.

[0074] In the example of FIG. 10, MODEL2 selects, as the optimal blending ratio pattern, the blending ratio pattern with the highest probability that "μ falls within the target range of 290 to 310" (hereinafter referred to as "probability with respect to the target range"). Then, MODEL2 outputs the optimal blending ratio pattern as the prediction result (i.e., prediction dataset 2300). In the example of FIG. 10, MODEL2 calculates the probability with respect to the target range for all combinations of the plurality of possible blending ratio patterns. Specifically, in the example of FIG. 10, MODEL2 calculates the probability with respect to the target range under the assumption that μ follows a normal distribution.

[0075] In the example of FIG. 10, a dataset showing the correspondence between "each blending ratio pattern", "μ and σ", and "probability with respect to the target range" is created for each id. In the example of FIG. 10, the dataset corresponding to the j-th id (idj) is referred to as DATASET2_idj.

[0076] In the example of FIG. 10, MODEL2 searches for the dataset with the maximum probability with respect to the target range (hereinafter referred to as the maximum probability dataset) for all j. In the example of FIG. 10, as a result of the search by MODEL2, it is found that DATASET2_id1 is the maximum probability dataset. Note that the first blending ratio pattern in the example of FIG. 10 is the blending ratio pattern of "x00 = 65, x01 = 5, x02 = 20, x03 = 0, x04 = 10".

[0077] MODEL2 selects the maximum probability dataset as the optimal dataset (DATASET2_OPT). Then, MODEL2 determines the blending ratio pattern corresponding to the optimal dataset as the optimal blending ratio pattern. In the example of FIG. 10, MODEL2 selects DATASET2_id1 as the optimal dataset. Then, the third algorithm execution unit 223 determines the above-mentioned first blending ratio pattern corresponding to DATASET2_id1 as the optimal blending ratio pattern. Thus, in the example of FIG. 10, the first blending ratio pattern is output as the prediction result.

[0078] As described above, the third algorithm execution unit 223 generates MODEL2 (a model that predicts a prediction data set 2300 that satisfies arbitrary resin composition characteristic data 2200) by the third algorithm (e.g., grid search). However, as is obvious to those skilled in the art, the method for determining the optimal blending ratio pattern is not limited to the above example.

[0079] For example, MODEL2 may search for a blending ratio pattern corresponding to a data set (hereinafter referred to as the prediction value closest data set) in which the prediction value (μ) closest to 300 (the median of the above-described numerical range of viscosity) is obtained among all the blending ratio patterns in the example of FIG. 10. Then, MODEL2 may determine the blending ratio pattern corresponding to the prediction value closest data set as the optimal blending ratio pattern.

[0080] (Example of derivation of recommended data in the recommended data derivation unit 34) By using MODEL2 generated as described above, in the recommended data derivation unit 34, recommended data 340 can be derived based on the third input data 330 (resin composition required characteristic data) (see also FIG. 4 above).

[0081] As an example, the third input data 330 may be data indicating a viscosity of "μ = 350 to 370". In this case, the recommended data derivation unit 34 can derive recommended data 340 corresponding to the third input data 330 by inputting the third input data 330 into MODEL2. The recommended data 340 in this example is, for example, an optimal blending ratio pattern corresponding to a viscosity of μ = 350 to 370.

[0082] (Effect) According to the model generation device 100, MODEL2 that predicts a prediction data set 2300 that satisfies arbitrary resin composition characteristic data 2200 can be generated. In other words, MODEL2 that outputs the prediction data set 2300 as a prediction result regarding the required characteristic satisfaction condition for the resin composition can be generated as a prediction model.

[0083] According to the prediction device 300, using MODEL2 previously generated by the model generation device 100, it is possible to predict the required property satisfaction conditions for the resin composition. Specifically, in the prediction device 300, by inputting the resin composition required property data into MODEL2, the recommended data 340 can be derived. As is clear from the above descriptions, the recommended data 340 is the prediction result for the resin composition required property data.

[0084] As described above, in the resin composition manufacturing system 1, (i) MODEL2 can be generated by machine learning, and (ii) using the MODEL2, it is possible to predict the required property satisfaction conditions for the resin composition. Therefore, according to the resin composition manufacturing system 1, different from the known technology (e.g., the technology of Patent Document 1), it is not necessary to consider the required property satisfaction conditions through a large number of trial and errors based on experiments. Therefore, it is possible to find the required property satisfaction conditions for the resin composition more efficiently than before.

[0085] Therefore, according to the resin composition manufacturing system 1, for example, a resin composition having desired required properties can be developed at a lower cost than before. Also, it is possible to more easily realize a new resin composition expected to contribute to the sustainable development goals (SDGs).

[0086] (Supplementary) By the way, "using machine learning to derive the recommended composition data of the resin composition" itself is known. However, as far as the inventors investigated, there is no literature that discloses or suggests "generating MODEL2 (a model for predicting the required property satisfaction conditions for the resin composition C containing the inorganic filler A and the resin B) based on the specific combination of each input data disclosed in Embodiment 1". Therefore, it can be said that the resin composition manufacturing system 1 (more specifically, the model generation device 100 and the prediction device 300) is based on a novel technical idea that is sufficiently differentiated from the prior art.

[0087] 〔Embodiment 2〕 In Embodiment 1, as an example of the first algorithm, weighted average calculation (an algorithm for calculating the weighted average of feature amounts) and principal component analysis (an algorithm for reducing the dimensions of feature amounts) were exemplified. However, as is obvious to those skilled in the art, the first algorithm is not limited to these algorithms.

[0088] For example, the first algorithm may be an n-th moment calculation algorithm (an algorithm for calculating the n-th moment of feature amounts). n is an arbitrary natural number. Note that the weighted average calculation algorithm in Embodiment 1 corresponds to the n-th moment calculation algorithm when n = 1. Alternatively, the first algorithm may be an algorithm for calculating known statistical amounts (e.g., average value, variance, maximum value, minimum value, etc.) for feature amounts.

[0089] [Example of Realization by Software] The functions of the resin composition manufacturing system 1 (hereinafter referred to as the "system") are programs for causing a computer to function as the system, and can be realized by programs for causing a computer to function as each control block of the system (particularly, each part included in the model generation device 100 and the prediction device 300).

[0090] In this case, the system includes a computer having at least one control device (e.g., a processor) and at least one storage device (e.g., a memory) as hardware for executing the above program. By executing the above program with this control device and storage device, each function described in each of the above embodiments is realized.

[0091] The above program may be recorded on one or more computer-readable recording media, not temporarily. This recording medium may or may not be provided in the above system. In the latter case, the above program may be supplied to the above system via any wired or wireless transmission medium.

[0092] In addition, part or all of the functions of each of the above control blocks can also be realized by a logic circuit. For example, an integrated circuit in which a logic circuit functioning as each of the above control blocks is formed is also included in the scope of one aspect of the present invention. In addition to this, for example, it is also possible to realize the functions of each of the above control blocks by a quantum computer.

[0093] Also, as described above, each process described in each of the above embodiments may be executed by AI (Artificial Intelligence). In this case, the AI may operate in the above control device, or may operate in another device (for example, an edge computer or a cloud server, etc.).

[0094] 〔Supplementary Notes〕 One aspect of the present invention is not limited to each of the above-described embodiments, and various modifications are possible within the scope shown in the claims. Embodiments obtained by appropriately combining the technical means disclosed in different embodiments are also included in the technical scope of one aspect of the present invention.

Explanation of Reference Numerals

[0095] 1 Resin Composition Manufacturing System 11 First Input Data Acquisition Unit 12 Second Input Data Acquisition Unit 21 First Machine Learning Unit 22 Second Machine Learning Unit 33 Third Input Data Acquisition Unit 34 Recommended Data Derivation Unit 35 Output Unit 100 Model Generation Device 110 First Input Data 111 Inorganic Filler Property Input Data 112 Resin Property Input Data 113 Inorganic Filler Blending Input Data 114 Resin Blending Input Data 120 Second Input Data (Resin Composition Property Input Data) 211 First Algorithm Execution Unit 212 Second Algorithm Execution Unit 223 Third algorithm execution unit 300 Prediction device 330 Third input data (resin composition required property data) 340 Recommended data 341 Recommended inorganic filler property data 342 Recommended resin property data 343 Recommended inorganic filler blending data 344 Recommended resin blending data 1100 Arbitrary input data set 1110 Arbitrary inorganic filler property data 1120 Arbitrary resin property data 1130 Arbitrary inorganic filler blending data 1140 Arbitrary resin blending data 1200 Unknown resin composition property data 2200 Arbitrary resin composition property data 2300 Prediction data set 2310 Predicted inorganic filler property data 2330 Predicted inorganic filler blending data 2340 Predicted resin blending data MODEL1 First prediction model MODEL2 Second prediction model X Explanatory variable y Objective variable

Claims

1. A model generation device that generates a prediction model for predicting one or more of (i) the properties of the inorganic filler, (ii) the properties of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, which satisfy the required properties of a resin composition containing one or more inorganic fillers and one or more resins, a first input data acquisition unit that acquires first input data, which is input data including at least one of (i) inorganic filler property input data indicating the properties of the inorganic filler, (ii) resin property input data indicating the properties of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the inorganic filler in the resin composition in which a plurality of types of the inorganic filler are blended with the resin, and (iv) resin blending input data regarding the blending ratio of the resin, a second input data acquisition unit that acquires resin composition property input data indicating the properties of the resin composition as second input data paired with the first input data, a first machine learning unit that generates a first prediction model for predicting unknown resin composition property data from one or more of (i) arbitrary inorganic filler property data, (ii) arbitrary resin property data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data, based on the first input data and the second input data, a second machine learning unit that generates a second prediction model for predicting one or more of (i) predicted inorganic filler property data, (ii) predicted resin property data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data, which satisfy arbitrary resin composition property data, based on the first prediction model, and is provided with, the first machine learning unit includes, a first algorithm execution unit that executes a first algorithm for deriving an explanatory variable corresponding to the data indicating the properties of the resin composition obtained from the second input data, a second algorithm execution unit that executes a second algorithm for generating the first prediction model based on the explanatory variable and the second input data, and has, the second algorithm is, at least one of Gaussian process regression, support vector machine, linear regression, decision tree, random forest, neural network, and gradient boosting tree, The second machine learning unit has a third algorithm execution unit that executes a third algorithm for generating the second prediction model based on the first prediction model. The third algorithm is at least one of a genetic algorithm, the method of steepest descent, grid search, and Bayesian optimization. Model generation device.

2. The inorganic filler property input data is of the inorganic filler composition formula, crystallinity, specific gravity, bulk density, particle size distribution, specific surface area, pore volume, zeta potential, electrical conductivity, dielectric constant, dielectric loss tangent, refractive index, specific heat, thermal conductivity, linear expansion coefficient, crushing strength, sphericity, aspect ratio, moisture content, carbon content, nitrogen content, surface functional group type, surface functional group amount, light absorption wavelength, absorbance, M value, and solubility parameter indicating at least one of them as the property of the inorganic filler. The model generation device according to claim 1.

3. The resin property input data is of the resin composition formula, degree of polymerization, molecular weight distribution, stereoregularity, reactive functional group type, reactive functional group amount, viscosity, melting point, glass transition temperature, crystallinity, elastic modulus, yield stress, tensile strength, fracture toughness, light absorption wavelength, absorbance, specific gravity, refractive index, electrical conductivity, dielectric constant, dielectric loss tangent, specific heat, thermal conductivity, moisture content, and solubility parameter indicating at least one of them as the property of the resin. The model generation device according to claim 1 or 2.

4. The resin composition property input data is of the resin composition viscosity, fluidity, moldability, adhesiveness, transparency, color tone, strength, water absorption rate, linear expansion coefficient, elastic modulus, yield stress, tensile strength, fracture toughness, electrical conductivity, dielectric constant, dielectric loss tangent, thermal conductivity, and stability indicating at least one of them as the property of the resin composition. The model generation device according to any one of claims 1 to 3.

5. Using the prediction model generated by a model generation device that generates a prediction model for predicting one or more of (i) the properties of the inorganic filler, (ii) the properties of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, to satisfy the required properties of a resin composition containing one or more inorganic fillers and one or more resins. A prediction device that predicts one or more of (i) the properties of the inorganic filler, (ii) the properties of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, that satisfy the required properties of the resin composition. The model generation device executes a first input data acquisition step of acquiring first input data that includes at least one of (i) inorganic filler property input data indicating the properties of the inorganic filler, (ii) resin property input data indicating the properties of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the inorganic filler in the resin composition in which a plurality of types of the inorganic filler are blended with the resin, and (iv) resin blending input data regarding the blending ratio of the resin. The model generation device executes a second input data acquisition step of acquiring resin composition property input data indicating the properties of the resin composition as second input data paired with the first input data. The model generation device executes a first machine learning step of generating a first prediction model that predicts unknown resin composition property data from one or more of (i) arbitrary inorganic filler property data, (ii) arbitrary resin property data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data, based on the first input data and the second input data. The model generation device executes a second machine learning step of generating a second prediction model that predicts one or more of (i) predicted inorganic filler property data, (ii) predicted resin property data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data, that satisfy arbitrary resin composition property data, based on the first prediction model. The first machine learning step includes a first algorithm execution step of executing a first algorithm in which the model generation device derives an explanatory variable corresponding to data indicating the properties of the resin composition obtained from the second input data, and a second algorithm execution step of executing a second algorithm in which the model generation device generates the first prediction model based on the explanatory variable and the second input data. The second algorithm is at least one of Gaussian process regression, support vector machine, linear regression, decision tree, random forest, neural network, and gradient boosting tree. The second machine learning process is a third algorithm execution step in which the model generation device executes a third algorithm for generating the second prediction model based on the first prediction model, The third algorithm is at least one of a genetic algorithm, a steepest descent method, a grid search, and a Bayesian optimization, The prediction device is a third input data acquisition unit that acquires, as third input data, resin composition required characteristic data indicating the required characteristics of the resin composition, by inputting the resin composition required characteristic data into the second prediction model generated by the model generation device, (i) recommended inorganic filler characteristic data indicating the characteristics of the inorganic filler that satisfies the resin composition required characteristic data, (ii) recommended resin characteristic data indicating the characteristics of the resin that satisfies the resin composition required characteristic data, (iii) recommended inorganic filler blending data indicating the blending of the inorganic filler that satisfies the resin composition required characteristic data, and (iv) recommended resin blending data indicating the blending of the resin that satisfies the resin composition required characteristic data, and a recommended data derivation unit that derives recommended data including at least one of them. Prediction device.

6. The resin composition required characteristic data is of the resin composition viscosity, fluidity, moldability, adhesiveness, transparency, color tone, strength, water absorption rate, linear expansion coefficient, elastic modulus, yield stress, tensile strength, fracture toughness, electrical conductivity, dielectric constant, dielectric loss tangent, thermal conductivity, and stability indicating at least one of them as the required characteristics of the resin composition, The prediction device according to claim 5.

7. A model generation method in which a model generation device generates a prediction model for predicting one or more of (i) the characteristics of the inorganic filler, (ii) the characteristics of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin that satisfy the required characteristics of a resin composition containing one or more inorganic fillers and one or more resins. a first input data acquisition step in which the model generation device acquires first input data that is input data including at least one of (i) inorganic filler property input data indicating the properties of the inorganic filler, (ii) resin property input data indicating the properties of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the inorganic filler in the resin composition in which a plurality of types of the inorganic filler are blended with the resin, and (iv) resin blending input data regarding the blending ratio of the resin; a second input data acquisition step in which the model generation device acquires resin composition property input data indicating the properties of the resin composition as second input data paired with the first input data; a first machine learning step in which the model generation device generates a first prediction model for predicting unknown resin composition property data from one or more of (i) arbitrary inorganic filler property data, (ii) arbitrary resin property data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data, based on the first input data and the second input data; a second machine learning step in which the model generation device generates a second prediction model for predicting one or more of (i) predicted inorganic filler property data, (ii) predicted resin property data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data that satisfy arbitrary resin composition property data, based on the first prediction model; The first machine learning step includes a first algorithm execution step in which the model generation device executes a first algorithm for deriving an explanatory variable corresponding to data indicating the properties of the resin composition obtained from the second input data; a second algorithm execution step in which the model generation device executes a second algorithm for generating the first prediction model based on the explanatory variable and the second input data; The second algorithm is at least one of Gaussian process regression, support vector machine, linear regression, decision tree, random forest, neural network, and gradient boosting tree; The second machine learning step includes a third algorithm execution step in which the model generation device executes a third algorithm for generating the second prediction model based on the first prediction model; The third algorithm is A model generation method that is at least one of a genetic algorithm, the steepest descent method, grid search, and Bayesian optimization. Model generation method.

8. Using the prediction model generated by a model generation device that generates a prediction model for predicting one or more of (i) the characteristics of the inorganic filler, (ii) the characteristics of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, to satisfy the required characteristics of a resin composition containing one or more inorganic fillers and one or more resins. A prediction method in which a prediction device predicts one or more of (i) the characteristics of the inorganic filler, (ii) the characteristics of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin, to satisfy the required characteristics of the resin composition. The model generation device executes a first input data acquisition step of acquiring first input data, which is input data including at least one of (i) inorganic filler characteristic input data indicating the characteristics of the inorganic filler, (ii) resin characteristic input data indicating the characteristics of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the inorganic filler in the resin composition in which a plurality of types of the inorganic filler are blended with the resin, and (iv) resin blending input data regarding the blending ratio of the resin. The model generation device executes a second input data acquisition step of acquiring resin composition characteristic input data indicating the characteristics of the resin composition as second input data paired with the first input data. The model generation device executes a first machine learning step of generating a first prediction model for predicting unknown resin composition characteristic data based on the first input data and the second input data, from one or more of (i) arbitrary inorganic filler characteristic data, (ii) arbitrary resin characteristic data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data. The model generation device executes a second machine learning step of generating a second prediction model for predicting one or more of (i) predicted inorganic filler characteristic data, (ii) predicted resin characteristic data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data, based on the first prediction model, to satisfy arbitrary resin composition characteristic data. The first machine learning step includes A first algorithm execution step of executing a first algorithm for deriving an explanatory variable corresponding to data indicating the characteristics of the resin composition obtained from the second input data by the model generation device; A second algorithm execution step of executing a second algorithm for generating the first prediction model by the model generation device based on the explanatory variable and the second input data, and The second algorithm is At least one of Gaussian process regression, support vector machine, linear regression, decision tree, random forest, neural network, and gradient boosting tree, The second machine learning step is The model generation device includes a third algorithm execution step of executing a third algorithm for generating the second prediction model based on the first prediction model, The third algorithm is At least one of genetic algorithm, steepest descent method, grid search, and Bayesian optimization, The prediction method is A third input data acquisition step of the prediction device acquiring resin composition required characteristic data indicating the required characteristics of the resin composition as third input data; The prediction device inputs the resin composition required characteristic data into the second prediction model generated by the model generation device, thereby (i) recommended inorganic filler characteristic data indicating the characteristics of the inorganic filler that satisfies the resin composition required characteristic data, (ii) recommended resin characteristic data indicating the characteristics of the resin that satisfies the resin composition required characteristic data, (iii) recommended inorganic filler blending data indicating the blending of the inorganic filler that satisfies the resin composition required characteristic data, and (iv) recommended resin blending data indicating the blending of the resin that satisfies the resin composition required characteristic data, and a recommended data derivation step of deriving at least one of the recommended data including; Prediction method.

9. A model generation device that generates a prediction model for predicting one or more of (i) the characteristics of the inorganic filler, (ii) the characteristics of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin that satisfy the required characteristics of a resin composition containing one or more inorganic fillers and one or more resins; A resin composition manufacturing system comprising a prediction device that predicts one or more of (i) the properties of the inorganic filler, (ii) the properties of the resin, (iii) the blending ratio of the inorganic filler, and (iv) the blending ratio of the resin that satisfy the required properties of the resin composition using the prediction model generated by the model generation device. The model generation device A first input data acquisition unit that acquires first input data which is input data including at least one of (i) inorganic filler property input data indicating the properties of the inorganic filler, (ii) resin property input data indicating the properties of the resin, (iii) inorganic filler blending input data regarding the blending ratio of the inorganic filler in the resin composition in which a plurality of types of the inorganic filler are blended with the resin, and (iv) resin blending input data regarding the blending ratio of the resin. A second input data acquisition unit that acquires resin composition property input data indicating the properties of the resin composition as second input data paired with the first input data. A first machine learning unit that generates a first prediction model for predicting unknown resin composition property data from one or more of (i) arbitrary inorganic filler property data, (ii) arbitrary resin property data, (iii) arbitrary inorganic filler blending data, and (iv) arbitrary resin blending data based on the first input data and the second input data. A second machine learning unit that generates a second prediction model for predicting one or more of (i) predicted inorganic filler property data, (ii) predicted resin property data, (iii) predicted inorganic filler blending data, and (iv) predicted resin blending data that satisfy arbitrary resin composition property data based on the first prediction model. The first machine learning unit A first algorithm execution unit that executes a first algorithm for deriving explanatory variables corresponding to data indicating the properties of the resin composition obtained from the second input data. A second algorithm execution unit that executes a second algorithm for generating the first prediction model based on the explanatory variables and the second input data. The second algorithm is at least one of Gaussian process regression, support vector machine, linear regression, decision tree, random forest, neural network, and gradient boosting tree. The second machine learning unit includes a third algorithm execution unit that executes a third algorithm for generating the second prediction model based on the first prediction model. The third algorithm is at least one of a genetic algorithm, a steepest descent method, a grid search, and Bayesian optimization. The prediction device includes a third input data acquisition unit that acquires, as third input data, resin composition required characteristic data indicating the required characteristics of the resin composition, and a recommended data derivation unit that derives recommended data including at least one of (i) recommended inorganic filler characteristic data indicating the characteristics of an inorganic filler that satisfies the resin composition required characteristic data, (ii) recommended resin characteristic data indicating the characteristics of a resin that satisfies the resin composition required characteristic data, (iii) recommended inorganic filler blending data indicating the blending of the inorganic filler that satisfies the resin composition required characteristic data, and (iv) recommended resin blending data indicating the blending of the resin that satisfies the resin composition required characteristic data, by inputting the resin composition required characteristic data into the second prediction model generated by the model generation device. Resin composition manufacturing system.

Images