Information processing method, information processing device, and program
The conversion of prediction formulas for raw material groups using constant adjustments addresses the versatility issue in existing models, enabling efficient and accurate predictions for new material compositions.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- SUMITOMO BAKELITE CO LTD
- Filing Date
- 2025-10-23
- Publication Date
- 2026-07-22
AI Technical Summary
Existing prediction models for material properties lack versatility and require significant effort to adapt to different raw material groups, with black-box machine learning models being difficult to apply and lacking interpretability.
A method and apparatus that convert a first formula for one raw material group into a second formula for a modified group by adjusting constants using the least squares method or Bayesian optimization, allowing repurposing of the first formula for new raw material groups with minimal data preparation.
Enhances the generality of prediction models by enabling efficient adaptation to new raw material groups with improved accuracy and interpretability, reducing the need for extensive data collection and model rebuilding.
Smart Images
Figure 0007893358000001_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to an information processing method, an information processing apparatus, and a program.
Background Art
[0002] In material design, attempts have been made to predict the properties of materials obtained from a plurality of raw materials.
[0003] Patent Document 1 describes symbolic regression based on genetic programming. It is described that symbolic regression can obtain a mathematical equation that gives the value of a dependent variable based on the input values of one or more independent variables.
Prior Art Documents
Patent Documents
[0004]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0005] However, in the technology of Patent Document 1, there is room for improvement from the viewpoint of the versatility of the prediction model.
[0006] One aspect of the present invention provides a technique for enhancing the versatility of a prediction model.
Means for Solving the Problems
[0007] According to one form of the present invention, the following information processing method, information processing apparatus, and program are provided.
[0008] (1) An information processing method executed by one or more computers, The method includes a conversion step of converting a first formula for calculating the material properties of a first material obtained using a first raw material group, based on multiple raw material property values for a first raw material group consisting of multiple raw materials, into a second formula for calculating the material properties of a second material obtained using a second raw material group, based on multiple raw material property values for a second raw material group. The second raw material group is a group of raw materials obtained by replacing at least one raw material from the first raw material group with another raw material. Information processing methods. (2) In the information processing method described in (1), In the conversion step, the first formula is converted to the second formula using a combination of multiple data sets of raw material properties for the second raw material group and material properties for the second material. Information processing methods. (3) In the information processing method described in (1) or (2), In the conversion step, the second formula is obtained by changing one or more constants in the first formula. Information processing methods. (4) In the information processing method described in (3), In the transformation step, the one or more constants are modified using the least squares method or Bayesian optimization. Information processing methods. (5) In the information processing method described in any one of (1) to (4), The evaluation step further includes evaluating the accuracy of at least one of the first and second formulas. Information processing methods. (6) An information processing method performed by one or more computers, The method includes a recommendation information generation step, which generates recommendation information regarding the values of the material properties of a first material obtained using a first raw material group, based on the response characteristics of the material properties to the raw material properties, in a first formula for calculating the values of the material properties of a first material obtained using the first raw material group from the values of multiple raw material properties of a first raw material group consisting of multiple raw materials. Information processing methods. (7) In the information processing method described in (6), The step further includes a response characteristic output step that outputs information indicating the aforementioned response characteristics. Information processing methods. (8) In the information processing method described in (6) or (7), If the response characteristics are linear, the recommendation information includes information indicating a first recommended value and a second recommended value that fall within the data acquisition range of the raw material characteristics. The difference between the first recommended value and the minimum value of the data acquisition range is smaller than the difference between the first recommended value and the central value of the data acquisition range. The difference between the second recommended value and the maximum value of the data acquisition range is smaller than the difference between the second recommended value and the central value of the data acquisition range. Information processing methods. (9) In the information processing method described in any one of (6) to (8), The process further includes a raw material characteristic acquisition step of acquiring the result of selecting one or more raw material characteristics from the aforementioned plurality of raw material characteristics, In the recommended information generation step, the recommended information is generated for each of the one or more selected raw material characteristics. Information processing methods. (10) In the information processing method described in any one of (6) to (8), In the recommendation information generation step, the recommendation information is generated for each of the multiple raw material characteristics. Information processing methods. (11) In the information processing method described in any one of (6) to (10), The information processing method further includes a material property acquisition step of acquiring the result of selecting one of several material properties, In the recommended information generation step, the recommended information is generated in the first formula for the selected material properties. Information processing methods. (12) In the information processing method described in any one of (6) to (11), The data points to be acquired are data points to be acquired in order to clarify the relationship between the raw material characteristics of the raw materials included in the second raw material group and the material characteristics of the second material obtained using the second raw material group. The second raw material group is a raw material group obtained by replacing at least one raw material in the first raw material group with another raw material. Information processing method. (13) In the information processing method according to any one of (1) to (12), The method further includes a specifying step of specifying the first mathematical formula using a plurality of data obtained by combining the values of the plurality of raw material characteristics of the first raw material group and the value of the material characteristic of the first material. Information processing method. (14) In the information processing method according to any one of (1) to (13), The plurality of raw material characteristics include a plurality of characteristics regarding one raw material. Information processing method. (15) A conversion unit that converts a first mathematical formula for calculating the value of the material characteristic of the first material obtained using the first raw material group from the values of the plurality of raw material characteristics of the first raw material group composed of a plurality of raw materials into a second mathematical formula for calculating the value of the material characteristic of the second material obtained using the second raw material group from the values of the plurality of raw material characteristics of the second raw material group, The second raw material group is a raw material group obtained by replacing at least one raw material in the first raw material group with another raw material. Information processing apparatus. (16) A computer is functioned as a conversion means for converting a first mathematical formula for calculating the value of the material characteristic of the first material obtained using the first raw material group from the values of the plurality of raw material characteristics of the first raw material group composed of a plurality of raw materials into a second mathematical formula for calculating the value of the material characteristic of the second material obtained using the second raw material group from the values of the plurality of raw material characteristics of the second raw material group, The second raw material group is a raw material group obtained by replacing at least one raw material in the first raw material group with another raw material. Program. Based on the response characteristics of the material characteristics with respect to the raw material characteristics in a first mathematical formula for calculating the value of the material characteristics of a first material obtained using the first raw material group from the values of a plurality of raw material characteristics of the first raw material group consisting of a plurality of raw materials, a recommendation information generation unit that generates recommendation information regarding the value of the raw material characteristics of the data points to be acquired is provided. Information processing apparatus. (18) A computer is Function as a recommendation information generation means for generating recommendation information regarding the value of the raw material characteristics of the data points to be acquired based on the response characteristics of the material characteristics with respect to the raw material characteristics in a first mathematical formula for calculating the value of the material characteristics of a first material obtained using the first raw material group from the values of a plurality of raw material characteristics of the first raw material group consisting of a plurality of raw materials. Program. [Effect of the Invention]
[0009] According to one aspect of the present invention, a technique for enhancing the generality of a prediction model can be provided. [Brief Description of the Drawings]
[0010] [Figure 1] It is a diagram showing an outline of an information processing apparatus according to the first embodiment. [Figure 2] It is a diagram exemplifying the relationship between a first mathematical formula and a second mathematical formula. [Figure 3] It is a diagram showing an outline of an information processing method according to the first embodiment. [Figure 4] It is a diagram exemplifying the functional configuration of an information processing apparatus according to the first embodiment. [Figure 5] It is a diagram exemplifying the flow of an information processing method according to the first embodiment. [Figure 6] It is a diagram exemplifying a computer for realizing an information processing apparatus. [Figure 7] It is a diagram showing an outline of an information processing apparatus according to the second embodiment. [Figure 8] It is a diagram showing an outline of an information processing method according to the second embodiment. [Figure 9]This figure shows an example of response characteristics. [Figure 10] This figure illustrates the functional configuration of an information processing device according to the second embodiment. [Figure 11] This diagram illustrates the flow of an information processing method according to the second embodiment. [Figure 12] This figure illustrates the functional configuration of an information processing device according to the third embodiment. [Figure 13] This diagram illustrates the flow of an information processing method according to the third embodiment. [Modes for carrying out the invention]
[0011] Embodiments of the present invention will be described below with reference to the drawings. In all drawings, similar components are denoted by the same reference numerals, and their descriptions are omitted as appropriate.
[0012] (First embodiment) Figure 1 is a diagram showing an overview of the information processing device 10 according to the first embodiment. Figure 2 is a diagram illustrating the relationship between the first formula and the second formula. The information processing device 10 includes a conversion unit 150. The conversion unit 150 converts the first formula to the second formula. The first formula is a formula for calculating the material properties of the first material from the values of multiple raw material properties for the first raw material group. The first raw material group consists of multiple raw materials. The first material is a material obtained using the first raw material group. The second formula is a formula for calculating the material properties of the second material from the values of multiple raw material properties for the second raw material group. The second material is a material obtained using the second raw material group. The second raw material group is a group of raw materials obtained by replacing at least one raw material from the first raw material group with another raw material.
[0013] Figure 3 is a diagram illustrating an overview of the information processing method according to this embodiment. The information processing method according to this embodiment is executed by one or more computers. The information processing method according to this embodiment includes a conversion step S10. In the conversion step S10, one or more computers convert a first formula into a second formula. The first formula is a formula for calculating the material properties of a first material from multiple raw material property values for a first raw material group. The first raw material group consists of multiple raw materials. The first material is a material obtained using the first raw material group. The second formula is a formula for calculating the material properties of a second material from multiple raw material property values for a second raw material group. The second material is a material obtained using the second raw material group. The second raw material group is a group of raw materials obtained by replacing at least one raw material from the first raw material group with another raw material.
[0014] The information processing method according to this embodiment may be executed by the information processing device 10.
[0015] The first formula can be described as a model for the first raw material group that predicts the characteristic values of the material obtained from the first raw material group. The second formula can be described as a model for the second raw material group that predicts the characteristic values of the material obtained from the second raw material group. According to the information processing device 10 and information processing method of this embodiment, the first formula for the first raw material group is converted into the second formula for the second raw material group. Therefore, the first formula can be repurposed for predictions concerning raw material groups other than the first raw material group. Furthermore, the second formula can be obtained with less effort compared to creating a new formula for the second raw material group from scratch. The information processing device 10 and information processing method of this embodiment will be described in detail below.
[0016] The first material obtained using the first raw material group and the second material obtained using the second raw material group may each be a composition such as a resin composition. The first material and the second material may each be a liquid or a solid.
[0017] The first material can be produced using all the raw materials included in the first raw material group. The first material may be obtained, for example, by mixing the raw materials included in the first raw material group, and may be obtained by further processing such as heating as needed.
[0018] The second material can be produced using all the raw materials included in the second raw material group. The second material may be obtained, for example, by mixing the raw materials included in the second raw material group, and may be obtained by further processing such as heating as needed.
[0019] Each raw material included in the first raw material group and each raw material included in the second raw material group are not particularly limited and may be inorganic or organic. Examples of inorganic materials include metals. A raw material may be, for example, any monomer, polymer, oligomer, or any substance as an additive. Examples of additives include curing agents, fillers, plasticizers, flame retardants, pigments, and dyes. A raw material may be a specific substance (e.g., silica, titanium dioxide, etc.), a composition, or a mixture. Raw materials may be identified by their substance name, product model number, etc. In the example in Figure 2, the first raw material group includes three raw materials: raw material A, raw material B, and raw material C, but the number of raw materials included in the first raw material group is not particularly limited. The number of raw materials included in the second raw material group is the same as the number of raw materials included in the first raw material group.
[0020] As described above, the second raw material group is a raw material group obtained by replacing at least one raw material from the first raw material group with another raw material. Replacing one raw material with another raw material is equivalent to, for example, replacing "silica" as a filler with "titanium dioxide". The other raw material can be said to be a substitute raw material or a similar raw material. If the first raw material group contains one or more monomers, the second raw material group may be a raw material group obtained by replacing at least one monomer from the first raw material group with another monomer. If the first raw material group contains one or more additives, the second raw material group may be a raw material group obtained by replacing at least one additive from the first raw material group with another additive. That is, if the first raw material group contains one or more curing agents, the second raw material group may be a raw material group obtained by replacing at least one curing agent from the first raw material group with another curing agent. If the first raw material group contains one or more fillers, the second raw material group may be a raw material group obtained by replacing at least one filler from the first raw material group with another filler. The same applies to other additives.
[0021] The second group of raw materials may be a group of raw materials obtained by replacing at least one raw material from the first group of raw materials with a raw material having a chemical structure similar to that raw material. A chemical structure similar to a certain chemical structure is, for example, a structure in which at least one functional group of that chemical structure is replaced by another functional group, a structure in which at least one functional group of that chemical structure is attached in a different position, a structure in which at least one functional group of that chemical structure is removed, or a structure in which at least one functional group is added to that chemical structure.
[0022] The second raw material group may be a group of raw materials in which at least one raw material from the first raw material group is replaced with a raw material having similar properties. A raw material having similar properties to a certain raw material is, for example, a raw material having a characteristic value of 90% to 110% of the characteristic value of the raw material for a specific raw material characteristic.
[0023] Raw material properties include, for example, hardness, strength, elastic modulus, melting point, boiling point, glass transition temperature, coefficient of thermal expansion, specific heat, heat capacity, thermal decomposition temperature, thermal diffusivity, thermal conductivity, light transmittance, reflectance, refractive index, absorbance, solubility, oxidation-reduction potential, acid resistance, alkali resistance, particle size, specific surface area, electrical resistivity, dielectric constant, magnetic permeability, and specific gravity. However, raw material properties are not limited to these examples.
[0024] Material properties include, for example, viscosity, hardness, strength, elastic modulus, melting point, boiling point, glass transition temperature, thermal expansion coefficient, specific heat, heat capacity, thermal decomposition temperature, thermal diffusivity, thermal conductivity, light transmittance, reflectance, refractive index, absorbance, solubility, oxidation-reduction potential, acid resistance, alkali resistance, particle size, specific surface area, electrical resistivity, dielectric constant, magnetic permeability, and specific gravity. However, material properties are not limited to these examples.
[0025] Even with the same raw materials, variations and variability can exist in their properties. The first formula is used to calculate the material properties of the first material from the values of multiple raw material properties for the first raw material group. In other words, the first formula is a regression equation in which multiple raw material properties for the first raw material group are explanatory variables and the material properties of the first material are the dependent variable. The multiple explanatory variables in the first formula may include the raw material properties of each of the multiple raw materials included in the first raw material group. Here, the raw material properties used as explanatory variables may differ for each raw material. For example, the multiple raw material properties used as explanatory variables may include the density of epoxy resin and the thermal conductivity of silica.
[0026] In one example, the multiple raw material properties used as explanatory variables may include multiple properties relating to a single raw material. For instance, the multiple raw material properties used as explanatory variables may include thermal conductivity, particle size, and density for a raw material called silica.
[0027] The second formula is used to calculate the material properties of the second material from the values of multiple material properties for the second raw material group. In other words, the second formula is a regression equation in which multiple material properties for the second raw material group are explanatory variables and the material properties of the second material are the dependent variable. The multiple explanatory variables in the second formula correspond to the multiple explanatory variables in the first formula. For example, suppose the second raw material group is a group of raw materials in the first raw material group in which raw material C is replaced with raw material C'. Furthermore, suppose that the density of raw material C is included as an explanatory variable in the first formula. In this case, the multiple explanatory variables in the second formula can be the same as the multiple explanatory variables in the first formula, but with the density of raw material C replaced by the density of raw material C'.
[0028] In conversion step S10, the conversion unit 150 obtains a second equation by changing one or more constants in the first equation. The constants may be in any position in the first equation. For example, the constants may be coefficients or exponents in the first equation. To improve prediction accuracy, the conversion unit 150 preferably obtains a second equation by changing multiple constants in the first equation.
[0029] In conversion step S10, the conversion unit 150 can convert the first formula to the second formula using multiple sets of data (hereinafter also referred to as "conversion data") that combine multiple raw material property values for the second raw material group with material property values for the second material. Multiple sets of conversion data can be prepared as a prior test by generating the second material using multiple second raw material groups with different raw material property values and measuring the material properties. Alternatively, multiple sets of conversion data may be prepared by simulation.
[0030] The conversion unit 150 obtains material property values (hereinafter also called "predicted values") by inputting the values of multiple material properties for the second group of raw materials included in the conversion data into the first equation. Here, the material properties of the replaced raw material (included in the first group of raw materials) in the first equation are replaced by the material properties of the replaced raw material (included in the second group of raw materials). The conversion unit 150 then compares the obtained predicted values with the material properties of the second material included in the conversion data and modifies one or more constants included in the first equation so that the difference between the predicted values and the material properties of the second material becomes small. The conversion unit 150 obtains the second equation by repeating this modification of constants using multiple conversion data. The conversion unit 150 can use, for example, the least squares method or Bayesian optimization to modify one or more constants.
[0031] Predictive models obtained through black-box machine learning are difficult to apply to different raw material groups. Therefore, when attempting to predict for a new raw material group, it is necessary to prepare a large amount of data again and generate a new model. Furthermore, there is a problem with the low interpretability of the predictive models, making it difficult to translate them into applicable design insights.
[0032] In contrast, by using mathematical formulas as a model to predict the properties of materials obtained from a certain group of raw materials, a common set of functions can be repurposed, and a predictive model can be obtained for new groups of raw materials with only a small amount of data preparation and constant adjustments. Furthermore, by utilizing mathematical formulas, it is possible to estimate the raw material properties that have a high contribution to the material properties.
[0033] Figure 4 is a diagram illustrating the functional configuration of the information processing device 10 according to this embodiment. In the example in Figure 4, the information processing device 10 further comprises an acquisition unit 110, a identification unit 130, an evaluation unit 170, and an output unit 190. Figure 5 is a diagram illustrating the flow of the information processing method according to this embodiment. However, the information processing device 10 and the information processing method according to this embodiment are not limited to this example, and it is sufficient that at least the first mathematical formula is converted to the second mathematical formula.
[0034] In step S101, the acquisition unit 110 acquires multiple data sets for the first formula. The data sets for the first formula are a combination of multiple raw material property values for the first raw material group and material property values for the first material. Multiple data sets for the first formula can be prepared as a preliminary test by generating the first material using multiple first raw material groups with different raw material property values and measuring the material properties. Alternatively, multiple data sets for the first formula may be prepared by simulation.
[0035] The acquisition unit 110 may acquire the first formula data input to the information processing device 10 by the user, or it may acquire the first formula data from a device other than the information processing device 10. Alternatively, the acquisition unit 110 may read and acquire the first formula data held in a storage unit accessible from the acquisition unit 110.
[0036] The information processing method according to this embodiment further includes a specific step S102. In the specific step S102, the specific unit 130 identifies a first mathematical formula using a plurality of data sets that combine multiple raw material characteristic values for the first raw material group and material characteristic values for the first material.
[0037] The identification unit 130 can identify the first formula using a symbolic regression method with multiple data for the first formula acquired by the acquisition unit 110.
[0038] Next, in step S103, the acquisition unit 110 acquires a plurality of conversion data. The acquisition unit 110 may acquire conversion data input to the information processing device 10 by the user, or it may acquire conversion data from a device other than the information processing device 10. Alternatively, the acquisition unit 110 may read and acquire conversion data held in a storage unit accessible from the acquisition unit 110.
[0039] In step S104, the conversion unit 150 uses the multiple conversion data acquired by the acquisition unit 110 to convert the first formula identified by the identification unit 130 into the second formula. Step S104 corresponds to the conversion step S10 described above. The method by which the conversion unit 150 converts the first formula into the second formula is as described above. The number of conversion data used to convert from the first formula to the second formula may be less than the number of first formula data used to identify the first formula.
[0040] If the information processing device 10 does not have a specific unit 130, the acquisition unit 110 may acquire the first mathematical formula, and the conversion unit 150 may use multiple conversion data to convert the first mathematical formula acquired by the acquisition unit 110 into a second mathematical formula. The acquisition unit 110 may acquire the first mathematical formula that has been input to the information processing device 10 by a user, or it may acquire the first mathematical formula from a device other than the information processing device 10. Alternatively, the acquisition unit 110 may read and acquire the first mathematical formula held in a storage unit accessible from the acquisition unit 110.
[0041] In the example shown in Figure 5, the information processing method further includes an evaluation step S105. In evaluation step S105, the evaluation unit 170 evaluates the accuracy of the second formula.
[0042] The evaluation unit 170 may also evaluate the accuracy of the first formula. The evaluation step S105 for evaluating the accuracy of the first formula may be performed before step S104. Alternatively, the evaluation unit 170 may evaluate the accuracy of both the first and second formulas. That is, the evaluation unit 170 evaluates the accuracy of at least one of the first and second formulas.
[0043] The evaluation unit 170 calculates, for example, the coefficient of determination of the formula as a predictive model as an evaluation metric. The method by which the evaluation unit 170 evaluates the formula is not limited to this example, and the formula can be evaluated using any existing method. Other examples of evaluation metrics include the mean absolute error (MAE) and the root mean square error (RMSE). The evaluation unit 170 may also perform cross-validation and calculate the standard deviation or coefficient of variation of the coefficient of determination, MAE, and RMSE as evaluation metrics. In this way, the stability of the predictive model can be evaluated.
[0044] In step S106, the output unit 190 outputs output data including at least the second mathematical formula. The destination of the output data is not particularly limited, but the output unit 190 may, for example, display the contents of the output data on a display connected to the information processing device 10, or transmit the output data to another device. Alternatively, the output unit 190 may store the output data in a storage unit accessible from the output unit 190.
[0045] If the information processing device 10 includes a specification unit 130, the output data may include the first mathematical formula identified by the specification unit 130. This allows the user to understand the first mathematical formula that formed the basis of the second mathematical formula.
[0046] If the information processing device 10 includes an evaluation unit 170, the output data may include an evaluation index of at least one of the first and second formulas calculated by the evaluation unit 170. This allows the user to understand the accuracy of each formula.
[0047] For example, if the evaluation result of the evaluation unit 170 for the first formula does not meet a predetermined standard, the user may prepare additional data for the first formula and have the identification unit 130 perform further symbolic regression processing to improve the accuracy of the first formula.
[0048] For example, if the evaluation result of the evaluation unit 170 for the second formula does not meet a predetermined standard, the user may prepare additional conversion data and have the conversion unit 150 further perform a process to change one or more constants to improve the accuracy of the second formula.
[0049] The hardware configuration of the information processing device 10 is described below. Each functional component of the information processing device 10 (acquisition unit 110, identification unit 130, conversion unit 150, evaluation unit 170, and output unit 190) is realized by a combination of hardware and software (e.g., a combination of an electronic circuit and a program to control it).
[0050] Figure 6 illustrates a computer 1000 for implementing the information processing device 10. Computer 1000 is any computer. For example, computer 1000 could be an SoC (System on Chip), a Personal Computer (PC), a server machine, a tablet terminal, or a smartphone. Computer 1000 may be a dedicated computer designed to implement the information processing device 10, or it may be a general-purpose computer. Furthermore, the information processing device 10 may be implemented by a single computer 1000, or by a combination of multiple computers 1000.
[0051] Computer 1000 includes a bus 1020, a processor 1040, memory 1060, a storage device 1080, an input / output interface 1100, and a network interface 1120. Bus 1020 is a data transmission path for the processor 1040, memory 1060, storage device 1080, input / output interface 1100, and network interface 1120 to send and receive data to and from each other. However, the method of connecting the processor 1040 and the other components is not limited to bus connection. Examples of the processor 1040 include various processors such as a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an FPGA (Field-Programmable Gate Array). Memory 1060 is a main memory device implemented using RAM (Random Access Memory), etc. Storage device 1080 is an auxiliary storage device implemented using a hard disk, SSD (Solid State Drive), memory card, or ROM (Read Only Memory), etc.
[0052] The input / output interface 1100 is an interface for connecting the computer 1000 with input / output devices. For example, input devices such as a keyboard and output devices such as a display are connected to the input / output interface 1100. The method by which the input / output interface 1100 connects to the input and output devices may be wireless or wired.
[0053] The network interface 1120 is an interface for connecting the computer 1000 to a network. Examples of such networks include LANs (Local Area Networks) and WANs (Wide Area Networks). The network interface 1120 may connect to the network via a wireless connection or a wired connection.
[0054] The storage device 1080 stores program modules that realize each functional component of the information processing device 10. The processor 1040 reads these program modules into the memory 1060 and executes them to realize the functions corresponding to each program module.
[0055] Each of the above-mentioned storage units may be provided in the information processing device 10, or it may be a storage device provided outside the information processing device 10. If the storage unit is provided in the information processing device 10, the storage unit may be implemented, for example, by the storage device 1080 of the computer 1000.
[0056] According to this embodiment, the conversion unit 150 converts the first formula into the second formula. The first formula is a formula for calculating the material properties of the first material from the values of multiple material properties for the first raw material group. The second formula is a formula for calculating the material properties of the second material from the values of multiple raw material properties for the second raw material group. The second raw material group is a group of raw materials obtained by replacing at least one raw material in the first raw material group with another raw material. Therefore, the first formula, which is a prediction model for the first raw material group, can be repurposed as the second formula, which is a prediction model for the second raw material group, thereby increasing its versatility.
[0057] (Second embodiment) Figure 7 shows an overview of the information processing device 20 according to the second embodiment. The information processing device 20 according to this embodiment includes a recommendation information generation unit 250. The recommendation information generation unit 250 generates recommendation information regarding the values of the raw material properties of data points to be acquired, based on the response characteristics of the material properties to the raw material properties in the first mathematical formula. The first mathematical formula is a formula for calculating the value of the material properties of a first material from the values of multiple raw material properties for a first raw material group consisting of multiple raw materials. The first material is a material obtained using the first raw material group.
[0058] Figure 8 is a diagram illustrating an overview of the information processing method according to this embodiment. The information processing method according to this embodiment is executed by one or more computers. The information processing method according to this embodiment includes a recommendation information generation step S20. In the recommendation information generation step S20, one or more computers generate recommendation information regarding the values of the raw material properties of the data points to be acquired, based on the response characteristics of the material properties to the raw material properties in the first mathematical formula. The first mathematical formula is a formula for calculating the value of the material properties of a first material from the values of multiple raw material properties for a first raw material group consisting of multiple raw materials. The first material is a material obtained using the first raw material group.
[0059] The information processing method according to this embodiment may be executed by the information processing device 20.
[0060] In the information processing device 10 according to the first embodiment, conversion data is used to convert the first mathematical formula into the second mathematical formula. Here, in order to reduce the effort required for testing to acquire data, it is preferable to obtain a highly accurate second mathematical formula with a small amount of conversion data. To that end, it is important to appropriately select data points. The information processing device 20 according to this embodiment can show the user what kind of data points are preferable to acquire in order to obtain a highly accurate second mathematical formula with a small amount of conversion data. The information processing device 20 and the information processing method according to this embodiment will be described in detail below.
[0061] The first formula, the first raw material group, the multiple raw materials, the first material, the raw material properties, and the material properties are as described in the first embodiment.
[0062] As described above, the recommended information relates to data points of raw material properties that should be acquired. The data points to be acquired are those that should be acquired in order to clarify the relationship between the raw material properties of the raw materials included in the second raw material group and the material properties of the second material obtained using the second raw material group. The recommended information generation unit 250 can generate recommended information based on the response characteristics of the material properties to the raw material properties in the first equation. Response characteristics refer to how the material properties calculated in the first equation change in response to a change in a certain raw material property. Specific examples of response characteristics include linear response, response with extreme values, converging response, and diverging response. It can also be said that the recommended information generation unit 250 generates recommended information based on the type of response characteristics.
[0063] Figure 9 shows an example of response characteristics. In the example in Figure 9, the material property values show a convergent response for raw material property 1, a response with extreme values for raw material property 2, and a linear response for raw material property 3. Thus, the response characteristics are specified for each raw material property.
[0064] The recommendation information generation unit 250 can identify the response characteristics of the first equation, for example, as follows. However, the method by which the recommendation information generation unit 250 identifies the response characteristics is not limited to the following example.
[0065] For each of the multiple raw material characteristics of the first raw material group, that is, the multiple explanatory variables included in the first formula, a reference value is predetermined. The recommendation information generation unit 250 then substitutes the reference value for all explanatory variables in the first formula except for the target explanatory variable. The target explanatory variable corresponds to the raw material characteristic for which the response characteristics are to be identified.
[0066] The recommendation information generation unit 250 then substitutes a predetermined set of values into the target explanatory variable and derives material property values for each value. In this way, the recommendation information generation unit 250 can obtain a set of data points (hereinafter referred to as "response data points") that show the relationship between the target explanatory variable and the material property.
[0067] The recommendation information generation unit 250 can further determine whether the response characteristics are linear based on these multiple response data points. For example, the recommendation information generation unit 250 can use these response data points to determine the correlation coefficient or the coefficient of determination R 2 The recommendation information generation unit 250 calculates the coefficient of determination R. 2 If the predetermined criteria are met (for example, if the difference between the coefficient of determination and 1 is less than or equal to a predetermined value), the response characteristic to the explanatory variable in question is identified as a linear response.
[0068] Furthermore, the recommendation information generation unit 250 can determine whether a response characteristic has an extremum based on multiple response data points. For example, the recommendation information generation unit 250 differentiates the material properties with respect to the relevant explanatory variable based on these response data points. Then, if the obtained derivative crosses zero as the explanatory variable changes, the recommendation information generation unit 250 determines that the response characteristic with respect to the relevant explanatory variable has an extremum.
[0069] The recommendation information generation unit 250 can determine whether or not the response characteristics converge based on multiple response data points. For example, the recommendation information generation unit 250 differentiates the material properties with respect to the target explanatory variable based on these response data points. Then, if the absolute value of the obtained derivative is less than or equal to a predetermined threshold (i.e., close to zero) within a range where the value of the target explanatory variable is above a predetermined value, the recommendation information generation unit 250 determines that the response characteristics with respect to the target explanatory variable converge.
[0070] The recommendation information generation unit 250 can determine whether the response characteristics diverge based on multiple response data points. For example, the recommendation information generation unit 250 differentiates the material properties with respect to the explanatory variable in question based on these response data points. If the obtained derivative value is monotonically increasing with respect to the explanatory variable, the recommendation information generation unit 250 determines that the response characteristics with respect to the explanatory variable in question diverge.
[0071] The recommended information is explained in detail below. Recommended information is specified for each raw material characteristic. For each raw material characteristic, there is a limit to the range in which the value can be changed, that is, the range in which data points can be obtained. The range in which data points can be obtained is called the "data acquisition range" for that raw material characteristic. Recommended information can be said to be information about any value within the data acquisition range.
[0072] If the data acquisition range is predetermined as a specific numerical range, the recommendation information may indicate the specific value or range of the raw material characteristic corresponding to each data point to be acquired. Alternatively, the recommendation information may indicate the position of the raw material characteristic value corresponding to each data point to be acquired relative to the entire data acquisition range. In other words, the recommendation information may indicate the value corresponding to each data point to be acquired when the data acquisition range is normalized. For example, the minimum value of the data acquisition range can be normalized to 0 and the maximum value to 100. In this way, even if the data acquisition range is unknown, the recommendation information can indicate each data point within the data acquisition range. However, the recommendation information may also be a message indicating the data point of the raw material characteristic to be acquired.
[0073] The recommendation information generation unit 250 can generate recommendation information regarding raw material properties based on the response characteristics of those raw material properties. That is, the recommendation information can indicate the raw material property values or the location within the data acquisition range of data points that should be acquired to clarify the relationship between the raw material properties and material properties, based on the identified response characteristics. In this way, the user can preferentially acquire data points that contribute greatly to identifying the features of the second mathematical formula. The recommendation information can be said to be information that indicates the data points to be acquired. The number of data points to be acquired indicated by the recommendation information is not particularly limited, but may be 2 or more and 10 or less. In this way, it can contribute to more efficient data acquisition by the user. An example of how the recommendation information generation unit 250 generates recommendation information will be described in detail below. However, the method by which the recommendation information generation unit 250 generates recommendation information is not limited to the following example.
[0074] When the response characteristics to raw material properties are linear, it is preferable to acquire at least data points near the minimum and maximum values of the data acquisition range in order to clarify the relationship between the raw material properties and material properties. That is, when the response characteristics are linear, it is preferable that the recommendation information includes information indicating a first recommended value and a second recommended value that are included in the data acquisition range of the raw material properties. Here, the first recommended value is a value close to the minimum value. That is, the difference between the first recommended value and the minimum value of the data acquisition range is smaller than the difference between the first recommended value and the center value of the data acquisition range. In a scale normalized with the minimum value of the data acquisition range at 0 and the maximum value at 100, it is preferable that the first recommended value is a value between 0 and 10. The first recommended value may also be the minimum value of the data acquisition range. Furthermore, the second recommended value is a value close to the maximum value. That is, the difference between the second recommended value and the maximum value of the data acquisition range is smaller than the difference between the second recommended value and the center value of the data acquisition range. In a scale normalized with the minimum value of the data acquisition range at 0 and the maximum value at 100, it is preferable that the second recommended value is a value between 90 and 100. The second recommended value may be the maximum value within the data acquisition range. When the response characteristics are linear, it is preferable that the recommendation information generation unit 250 generates recommendation information that includes information indicating the first recommended value and the second recommended value. Also, when the response characteristics are linear, the recommendation information generation unit 250 may generate recommendation information indicating that multiple data points should be acquired at equal intervals within the data acquisition range.
[0075] If the response characteristics to the raw material characteristics are responses that have extreme values, it is preferable that the recommendation information generation unit 250 generates recommendation information that includes values near the point where the extreme value is predicted, and information indicating the first recommendation value and the second recommendation value described above. Here, the values near the point where the extreme value is predicted may be the values of the raw material characteristics that take extreme values at the above-mentioned plurality of response data points. That is, the point where the extreme value occurs may be predicted based on the first formula.
[0076] When the response characteristics to raw material properties are convergent responses, it is preferable to acquire at least data points in regions where the changes are large on a logarithmic or exponential scale in order to clarify the relationship between the raw material properties and material properties. The recommendation information generation unit 250 preferably generates recommendation information that includes the above-mentioned first recommendation value, second recommendation value, and information indicating at least three values that are equally spaced between the first recommendation value and the second recommendation value on a logarithmic or exponential scale.
[0077] Figure 10 is a diagram illustrating the functional configuration of the information processing device 20 according to this embodiment. In the example of Figure 10, the information processing device 20 further comprises an acquisition unit 210, a specification unit 230, and an output unit 290. Figure 11 is a diagram illustrating the flow of the information processing method according to this embodiment. However, the information processing device 20 and the information processing method according to this embodiment are not limited to this example, and it is sufficient that at least recommendation information is generated.
[0078] In the example shown in Figure 11, the information processing method includes a specific step S202. In the specific step S202, the specific unit 230 identifies a first mathematical formula using a plurality of data sets that combine multiple raw material characteristic values for the first raw material group and material characteristic values for the first material.
[0079] In step S201, the acquisition unit 210 performs the same processing as the acquisition unit 110 in step S101 as described in the first embodiment. In the identification step S202, the identification unit 230 performs the same processing as the identification unit 130 as described in the first embodiment.
[0080] In the identification step S202, the identification unit 230 may identify the first formula for each of the multiple material properties. Then, in the example in Figure 11, the information processing method further includes a material property acquisition step S203 in which the result of selecting one of the multiple material properties is obtained. Then, in the recommendation information generation step S205, which will be described later, the recommendation information generation unit 250 generates recommendation information for the first formula for the selected material property. In this way, the user can obtain recommendation information for the material property of interest.
[0081] In the material property acquisition step S203, for example, multiple material properties (e.g., elastic modulus, glass transition temperature, and dielectric constant) are displayed as options on a display connected to the information processing device 20. The user performs an operation on the information processing device 20 to select one of the material properties. The acquisition unit 210 then acquires information indicating the user's selection result (i.e., the selected material property).
[0082] However, the information processing method according to this embodiment does not necessarily have to include the material property acquisition step S203. For example, the recommendation information generation unit 250 may generate recommendation information for predetermined material properties. In another example, the identification unit 230 may identify the first formula only for the material properties selected in the material property acquisition step S203.
[0083] In the example shown in Figure 11, the information processing method includes a raw material characteristic acquisition step S204. In the raw material characteristic acquisition step S204, the acquisition unit 210 acquires the result of selecting one or more raw material characteristics from among multiple raw material characteristics. Then, in the recommendation information generation step S205, which will be described later, the recommendation information generation unit 250 generates recommendation information for each of the one or more selected raw material characteristics. In this way, the user can obtain recommendation information for the desired raw material characteristics.
[0084] In the raw material properties acquisition step S204, for example, a display connected to the information processing device 20 shows multiple raw material properties (e.g., density, particle size, and specific surface area) as explanatory variables for the first mathematical formula as options. The user performs an operation to select one or more raw material properties from the information processing device 20. The acquisition unit 210 then acquires information indicating the user's selection result (i.e., the one or more raw material properties that were selected).
[0085] However, the information processing method according to this embodiment does not necessarily have to include the raw material characteristic acquisition step S204. For example, the recommendation information generation unit 250 may generate recommendation information for predetermined raw material characteristics. Alternatively, in the recommendation information generation step S205, recommendation information may be generated for each of the multiple raw material characteristics that are explanatory variables of the first formula.
[0086] In the recommendation information generation step S205, the recommendation information generation unit 250 generates recommendation information regarding the target raw material properties and target material properties using the first formula as described above. The recommendation information generation step S205 corresponds to the recommendation information generation step S20 described above. The recommendation information generation unit 250 may generate recommendation information for each of several combinations of material properties and raw material properties according to the results of the material property acquisition step S203 and raw material property acquisition step S204 described above, or according to predetermined information.
[0087] If the information processing device 20 does not have a specific unit 230, the acquisition unit 210 may acquire the first mathematical formula, and the recommendation information generation unit 250 may generate recommendation information using the first mathematical formula acquired by the acquisition unit 210. The acquisition unit 210 may acquire the first mathematical formula that has been input to the information processing device 20 by the user, or it may acquire the first mathematical formula from a device other than the information processing device 20. Alternatively, the acquisition unit 210 may read and acquire the first mathematical formula held in a storage unit accessible from the acquisition unit 210.
[0088] In step S206, the output unit 290 outputs output data that includes at least recommended information. The destination of the output data is not particularly limited, but the output unit 290 may, for example, display the contents of the output data on a display connected to the information processing device 20, or transmit the output data to another device. Alternatively, the output unit 290 may store the output data in a storage unit accessible from the output unit 290.
[0089] When the output unit 290 displays the contents of the output data on a display connected to the information processing device 20, the display may show characters, numbers, or diagrams to help the user understand the raw material characteristics of the data points to be acquired.
[0090] If the information processing device 20 includes a specification unit 230, the output data may include the first mathematical formula identified by the specification unit 230. This allows the user to understand the first mathematical formula that formed the basis of the recommended information.
[0091] The output data output by the output unit 290 may include information indicating the response characteristics of the first equation, associated with the recommendation information. That is, the information processing method according to this embodiment may include a response characteristic output step S206 that outputs information indicating the response characteristics. In this way, the user can recognize how the values of the target raw material characteristics affect the target material characteristics.
[0092] The hardware configuration of the computer implementing the information processing device 20 according to this embodiment is shown, for example, in Figure 6, similar to the information processing device 10 according to the first embodiment. However, the storage device 1080 of the computer 1000 implementing the information processing device 20 stores program modules that implement each functional component of the information processing device 20 (acquisition unit 210, identification unit 230, recommendation information generation unit 250, and output unit 290).
[0093] Each of the above-mentioned storage units may be provided in the information processing device 20, or it may be a storage device provided outside the information processing device 20. If the storage unit is provided in the information processing device 20, the storage unit may be implemented, for example, by the storage device 1080 of the computer 1000 that implements the information processing device 20.
[0094] According to this embodiment, the same functions and effects as in the first embodiment can be obtained. In addition, appropriate data points are proposed for converting the first formula to the second formula. By preparing conversion data based on the proposal, the first formula, which is a prediction model for the first raw material group, can be efficiently repurposed as the second formula, which is a prediction model for the second raw material group, thereby increasing its versatility.
[0095] (Third embodiment) Figure 12 is a diagram illustrating the functional configuration of the information processing device 10 according to the third embodiment. Figure 13 is a diagram illustrating the flow of the information processing method according to this embodiment. The information processing device 10 according to this embodiment is the same as the information processing device 10 according to the first embodiment, except for the points described below. The information processing device 10 according to this embodiment combines the functions of the information processing device 10 according to the first embodiment and the information processing device 20 according to the second embodiment.
[0096] The information processing device 10 according to this embodiment includes a recommendation information generation unit 140. The recommendation information generation unit 140 has the same functions as the recommendation information generation unit 250 described in the second embodiment. The acquisition unit 110 and the output unit 190 each combine the functions of the acquisition unit 210 and the output unit 290 according to the second embodiment.
[0097] Steps S301, S302, S307, S308, S309, and S310 are the same as steps S101 to S106 in the first embodiment, respectively. Steps S303, S304, S305, and S306 are the same as steps S203 to S206 in the second embodiment, respectively.
[0098] The user can perform tests to acquire the data points that are recommended in the recommended information output in step S306. Then, in step S307, the acquisition unit 110 acquires the multiple data points obtained as multiple conversion data. In this way, the first formula can be efficiently converted into the second formula.
[0099] The hardware configuration of the computer implementing the information processing device 10 according to this embodiment is shown, for example, in Figure 6, similar to the information processing device 10 according to the first embodiment. However, the storage device 1080 of the computer 1000 implementing the information processing device 10 according to this embodiment also stores a program module that implements the recommendation information generation unit 140.
[0100] According to this embodiment, the same actions and effects as those of the first and second embodiments can be obtained.
[0101] The embodiments of the present invention have been described above with reference to the drawings, but these are merely examples of the present invention, and various other configurations can also be adopted.
[0102] Furthermore, while the flowcharts used in the above description show multiple steps (processes) in sequence, the execution order of the steps performed in each embodiment is not limited to the order in which they are described. In each embodiment, the order of the illustrated steps can be changed to the extent that it does not impede the content. Also, the above embodiments can be combined to the extent that their contents do not conflict. [Explanation of symbols]
[0103] 10 Information Processing Devices 20 Information Processing Devices 110 Acquisition Department 130 Specific section 140 Recommended Information Generation Unit 150 Conversion Unit 170 Evaluation Department 190 Output section 210 Acquisition Department 230 Specific section 250 Recommended Information Generation Unit 290 Output section 1000 calculator 1020 Bus 1040 processor 1060 memory 1080 Storage Devices 1100 Input / Output Interface 1120 Network Interface
Claims
1. An information processing method performed by one or more computers, The method includes a conversion step of converting a first formula for calculating the material properties of a first material obtained using a first raw material group, based on multiple raw material property values for a first raw material group consisting of multiple raw materials, into a second formula for calculating the material properties of a second material obtained using a second raw material group, based on multiple raw material property values for a second raw material group. The second raw material group is a group of raw materials obtained by replacing at least one raw material from the first raw material group with another raw material. The explanatory variables of the first formula are the multiple raw material characteristics for the first raw material group, The explanatory variables of the second formula are the multiple raw material characteristics for the second raw material group, In the conversion step, the second formula is obtained by changing one or more constants in the first formula. Information processing methods.
2. In the information processing method described in claim 1, In the conversion step, the first formula is converted to the second formula using a plurality of data sets that combine the values of the plurality of raw material properties for the second raw material group and the values of the material properties of the second material. Information processing methods.
3. In the information processing method according to claim 1 or 2, In the transformation step, the one or more constants are modified using the least squares method or Bayesian optimization. Information processing methods.
4. In the information processing method according to claim 1 or 2, The evaluation step further includes evaluating the accuracy of at least one of the first and second formulas. Information processing methods.
5. In the information processing method according to claim 1 or 2, The further step includes a specific step of identifying the first mathematical formula using a plurality of data sets that combine the values of the plurality of raw material properties for the first raw material group and the values of the material properties of the first material, Information processing methods.
6. In the information processing method according to claim 1 or 2, The aforementioned multiple raw material properties include multiple properties relating to a single raw material. Information processing methods.
7. In the information processing method according to claim 1 or 2, The second raw material group is a group of raw materials obtained by replacing at least one monomer from the first raw material group with another monomer, or a group of raw materials obtained by replacing at least one curing agent from the first raw material group with another curing agent, or a group of raw materials obtained by replacing at least one filler from the first raw material group with another filler. Information processing methods.
8. In the information processing method according to claim 1 or 2, The second group of raw materials is a group of raw materials obtained by replacing at least one of the raw materials from the first group of raw materials with a raw material having a structure in which at least one functional group in the chemical structure of that raw material is replaced by another functional group, a structure in which at least one functional group in the chemical structure of that raw material is in a different bonding position, a structure in which at least one functional group in the chemical structure of that raw material is removed, or a structure in which at least one functional group is added to the chemical structure of that raw material. Information processing methods.
9. In the information processing method according to claim 1 or 2, The second raw material group is a group of raw materials obtained by replacing at least one raw material from the first raw material group with a raw material having a characteristic value of 90% or more and 110% or less of the characteristic value of that raw material with respect to a specific raw material characteristic of that raw material. Information processing methods.
10. In the information processing method according to claim 1 or 2, The first equation is a regression equation in which the material properties of the first material are the dependent variable. The second formula is a regression equation in which the material properties of the second material are the dependent variable. Information processing methods.
11. The system includes a conversion unit that converts a first formula for calculating the material properties of a first material obtained using a first raw material group, based on multiple raw material property values for a first raw material group consisting of multiple raw materials, into a second formula for calculating the material properties of a second material obtained using a second raw material group, based on multiple raw material property values for a second raw material group. The second raw material group is a group of raw materials obtained by replacing at least one raw material from the first raw material group with another raw material. The explanatory variables of the first formula are the multiple raw material characteristics for the first raw material group, The explanatory variables of the second formula are the multiple raw material characteristics for the second raw material group, The conversion unit obtains the second equation by changing one or more constants in the first equation. Information processing device.
12. Computers, A first formula for calculating the material properties of a first material obtained using a first raw material group, based on multiple raw material property values for a first raw material group consisting of multiple raw materials, is used as a conversion means to convert a first formula for calculating the material properties of a second material obtained using a second raw material group, based on multiple raw material property values for a second raw material group. The second raw material group is a group of raw materials obtained by replacing at least one raw material from the first raw material group with another raw material. The explanatory variables of the first formula are the multiple raw material characteristics for the first raw material group, The explanatory variables of the second formula are the multiple raw material characteristics for the second raw material group, The conversion means obtains the second formula by changing one or more constants in the first formula. program.