Design support device, design support method, program, and information processing system

Abstract

A design assistance device that assists the design of a material to which a plurality of raw materials are blended using a machine learning model that learns a correspondence relationship between design condition information of the material to which the plurality of raw materials are blended and characteristic information of the material, the design assistance device comprising: a raw material information storage unit that stores the design condition information of the material to which the plurality of raw materials are blended and the characteristic information of the material; the storage unit is configured to store information of names and attributes of various registered raw materials; a registration accepting section configured to accept an input of information of a name and an attribute of a raw material to be newly registered, and store the information to the raw material information storage section; and a feature amount generation section configured to generate a feature amount of the raw material that can be input to the machine learning model, based on the information of the attributes of the plurality of types of raw materials that are selected as the design condition information and stored in the raw material information storage section.

Description

Technical Field

[0001] The present disclosure relates to a design assistance device, a design assistance method, a program, and an information processing system. Background Art

[0002] Traditionally, a material design assistance device is known, which uses a machine learning model that has learned an experimental data set (learning data set) to predict the properties of materials including multiple raw materials included in the learning data set, and the experimental data set (learning data set) records the correspondence between the material design condition information (the combination amount and process conditions of each raw material, etc.) and the material characteristic information.

[0003] For example, Patent Document 1 describes a material property prediction system for processing case data including a plurality of records consisting of material composition, experimental conditions, and material properties to predict material properties.

[0004] <Prior Art Literature>

[0005] <Patent Documents>

[0006] Patent Document 1: Japanese Patent Application Publication No. 2021-47627 Summary of the invention

[0007] <Problems to be Solved by the Invention>

[0008] It would be very convenient if the conventional material design support device could predict the material properties of raw materials not included in the blending learning data set to support the material design. It should be noted that Patent Document 1 does not describe this content.

[0009] The purpose of the present disclosure is to provide a design support device, a design support method, a program, and an information processing system, which predict the characteristics of materials of raw materials not included in a combination learning data set to support the design of materials.

[0010] <Methods used to solve the problem>

[0011] The present disclosure has the following aspects.

[0012] [1] A design assisting device, using a machine learning model to assist in the design of a material for combining multiple raw materials, wherein the machine learning model learns the correspondence between design condition information of the material for combining multiple raw materials and characteristic information of the materials, the design assisting device comprising:

[0013] A raw material information storage unit configured to store information on the names and properties of a plurality of registered raw materials;

[0014] a registration receiving unit configured to receive input of information on the name and attributes of a raw material to be newly registered, and store the information in the raw material information storage unit; and

[0015] The feature quantity generating unit is configured to generate feature quantities of the raw materials that can be input into the machine learning model based on information on properties of multiple raw materials selected as the design condition information and stored in the raw material information storage unit.

[0016] [2] The design support device according to [1], wherein the registration accepting unit is configured to prohibit new registration of a raw material that matches the information on the molecular structure of the raw material already stored in the raw material information storage unit.

[0017] [3] A design assistance device according to [1], wherein the information on the properties of the raw material includes information on the purpose of the raw material, and the registration acceptance unit is configured to prohibit new registration of a raw material that is consistent with the information on the molecular structure of the raw material and the purpose of the raw material already stored in the raw material information storage unit.

[0018] [4] The design support device according to [3], wherein the registration acceptance unit is configured to prohibit new registration of a material that matches the name of the material already stored in the material information storage unit.

[0019] [5] A design assistance device according to any one of [1] to [4], wherein the registration receiving unit is configured to receive input of information on the molecular structure of a raw material in any one of SMILES notation, SMARTS notation, or InChI notation as information on the properties of the raw material to be newly registered.

[0020] [6] A design assistance device according to any one of [1] to [4], wherein the registration receiving unit is configured to receive input of information on the molecular structure of a raw material in any one of MOL format, SDF format, PDB format, or CIF format as information on the properties of the raw material to be newly registered.

[0021] [7] The design support device according to any one of [1] to [6], wherein the feature quantity generating unit is configured to generate the feature quantity of the raw material based on information on the molecular structure of the raw material using a fingerprint method.

[0022] [8] The design support device according to any one of [1] to [7], wherein the feature quantity generating unit is configured to generate the feature quantity of the raw material by calculating a physical property value of a molecule based on a molecular structure of the raw material.

[0023] [9] The design support device according to any one of [1] to [8], wherein the information on the attribute inputted by the registration receiving unit includes a physical property value of the raw material.

[0024]

[10] The design assistance device according to any one of [1] to [9] comprises: a design condition input receiving unit, configured to receive input of the design condition information including information of multiple raw materials stored in the raw material information storage unit; and a material property prediction unit, configured to predict the properties of the material using the machine learning model based on the design condition information obtained by receiving the input.

[0025]

[11] The design assistance device according to any one of [1] to [9] further includes: a required characteristic input receiving unit, configured to receive input of required characteristic information of the material; a range information input receiving unit, configured to receive input of range information of the design condition information including information of a plurality of raw materials stored in the raw material information storage unit; and a design condition suggestion unit, configured to suggest candidates for the design condition information that meet the required characteristic information based on a result obtained by predicting the characteristics of the material based on the design condition information within the range of the range information using the machine learning model.

[0026]

[12] The design support device according to

[10] or

[11] , further comprising: a display control unit configured to display the predicted characteristics of the material or the recommended candidates of the design condition information.

[0027]

[13] A design assistance method, in which a computer uses a machine learning model to assist in the design of a material that is combined with multiple raw materials, wherein the machine learning model learns the correspondence between the design condition information of the material that is combined with multiple raw materials and the characteristic information of the material, and the design assistance method includes: a registration acceptance step, in which the input of the name and attribute information of the raw material to be newly registered is accepted, and the information is registered in a raw material information storage unit, wherein the raw material information storage unit stores the name and attribute information of the multiple registered raw materials; and a feature quantity generation step, in which the feature quantity of the raw material that can be input into the machine learning model is generated based on the attribute information of the multiple raw materials selected as the design condition information and stored in the raw material information storage unit.

[0028]

[14] A program for causing a computer to execute a process, wherein the computer uses a machine learning model to assist in the design of a material that is combined with multiple raw materials, and the machine learning model learns the correspondence between design condition information of a material that is combined with multiple raw materials and characteristic information of the material, and the process includes: a registration acceptance process, which accepts input of information on the name and attributes of a raw material to be newly registered and registers it in a raw material information storage unit, wherein the raw material information storage unit stores information on the names and attributes of multiple registered raw materials; and a feature quantity generation process, which generates a feature quantity of the raw material that can be input into the machine learning model based on information on the attributes of multiple raw materials selected as the design condition information and stored in the raw material information storage unit.

[0029]

[15] An information processing system, in which multiple computers use a machine learning model to assist in the design of a material that is combined with multiple raw materials, wherein the machine learning model learns the correspondence between design condition information of a material that is combined with multiple raw materials and characteristic information of the material, and the information processing system comprises: a raw material information storage unit, configured to store information on the names and attributes of multiple registered raw materials; a registration receiving unit, configured to receive input of information on the names and attributes of raw materials to be newly registered, and store it in the raw material information storage unit; and a feature quantity generating unit, configured to generate a feature quantity of the raw material that can be input into the machine learning model based on information on the attributes of the multiple raw materials selected as the design condition information and stored in the raw material information storage unit.

[0030] <Effects of the Invention>

[0031] According to the present disclosure, it is possible to provide a design support device, a design support method, a program, and an information processing system that predict the characteristics of materials of raw materials not included in a combination learning data set to support the design of materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] [ Figure 1 ] is a structural diagram of an example of an information processing system according to the present embodiment.

[0033] [ Figure 2 ] is a hardware configuration diagram of an example of a computer according to the present embodiment.

[0034] [ Figure 3 ] is a functional structure diagram of an example of an information processing system according to the present embodiment.

[0035] [ Figure 4 ] is a schematic diagram of an example of a new raw material registration screen according to the present embodiment.

[0036] [ Figure 5 ] is a schematic diagram of an example of a new raw material registration screen according to the present embodiment.

[0037] [ Figure 6 ] is a diagram showing an example of a registered raw material list screen.

[0038] [ Figure 7 ] is a flowchart showing an example of a process of prohibiting registration of a new raw material having an equivalent molecular structure.

[0039] [ Figure 8 ] is a flowchart showing an example of a process of prohibiting registration of a new raw material having an equivalent molecular structure.

[0040] [ Fig. 9 ] is a flowchart showing an example of a process of prohibiting registration of a new raw material having an equivalent molecular structure.

[0041] [ Fig.10 ] is a flowchart representing an example of a process of assisting material design based on forward problem analysis.

[0042] [ Fig.11A ] is a composition diagram of an example of design condition information of a material.

[0043] [ Fig. 11B ] is a composition diagram of an example of material property prediction information.

[0044] [ Fig.12 ] is an explanatory diagram of an example of a process for generating characteristic quantities of a raw material to be blended into a material.

[0045] [ Fig.13 ] is a flowchart representing an example of a process of material design assisted by reverse problem analysis.

[0046] [ Fig.14 ] is a composition diagram of an example of required characteristic information of a material.

[0047] [ Fig.15 ] is a composition diagram of an example of range information of material design condition information.

[0048] [ Fig.16A ] is a diagram showing an example of the structure of information of the suggested material.

[0049] [ Fig. 16B ] is a diagram showing an example of the structure of information of the suggested material. DETAILED DESCRIPTION

[0050] Next, embodiments of the present invention will be described in detail. However, the present invention is not limited to the following embodiments.

[0051] [First embodiment]

[0052] <System Configuration>

[0053] Figure 1 It is a structural diagram of an example of the information processing system according to the present embodiment. Figure 1 The information processing system 1 includes a design support apparatus 10 and a user terminal 12. The design support apparatus 10 and the user terminal 12 are connected to each other via a communication network 18 such as a local area network (LAN) or the Internet so as to be capable of data communication.

[0054] The user terminal 12 is an information processing terminal operated by an operator, such as a PC, a tablet terminal, or a smart phone. The user terminal 12 displays a screen for receiving information input from the operator on a display device, and receives information input from the operator. In addition, the user terminal 12 sends the information received from the operator to the design support device 10 so that it performs processing for assisting the design of a material for mixing a plurality of raw materials. The user terminal 12 receives information on the execution result of the processing of the design support device 10, and displays the information on the display device so that the operator can confirm it.

[0055] The design support device 10 is an information processing device such as a PC that supports the design of a material that is mixed with a plurality of raw materials in the following manner. The design support device 10 performs a reverse problem analysis process that suggests candidates for design condition information of a material that satisfies required characteristic information, and a forward problem analysis process that predicts the characteristics of the material based on the design condition information. The design support device 10 supports the design of a material that is mixed with a plurality of raw materials by performing the reverse problem analysis process and the forward problem analysis process.

[0056] The design support device 10 can use a mathematical model such as a machine learning model (a machine learning model that has been learned), which has learned the correspondence between the design condition information of a material that combines multiple raw materials and the characteristic information of the material. As a machine learning method, supervised learning methods such as linear, generalized linear (lasso, ridge, elastic net, logistic), partial least squares, kernel ridge, Gaussian process, K nearest neighbor method, decision tree, random forest, adaptive boosting (adaBoost), guided aggregation (bagging), gradient boosting, support vector machine, or neural network can be used.

[0057] In the process of reverse problem analysis, the range information of the required characteristic information and the design condition information is input, and an exhaustive search point is generated in the design condition information within the range at random or with a predetermined step size. In addition, in the process of reverse problem analysis, the characteristics of the material corresponding to each exhaustive search point are predicted by using the machine learning model that has been learned, and the exhaustive search points that meet the characteristics of the required characteristic information are extracted, so that candidates of the design condition information that meet the desired required characteristic information can be obtained.

[0058] When predicting the properties of a material containing multiple raw materials through a machine learning model that has been learned, the composition ratio or the amount of each raw material mixed in the material is input. The design support device 10 creates the characteristic value (explanatory variable) of the raw material to be input into the machine learning model that has been learned based on the information obtained by digitizing the molecular structure of the raw material, and the composition ratio and the amount of the raw material. The following is an example of the amount of raw materials mixed.

[0059] As an example of a method for digitizing the molecular structure of a raw material, there is Extended Circular FingerPrints (hereinafter referred to as ECFP). ECFP is a method for digitizing the characteristics of the molecular structure by extracting the types and numbers of all partial structures from the molecular structure of the raw material and expressing them as a vector (column: type, value: number).

[0060] The input of the molecular structure information of the raw material can be performed using any of the notation methods of Simplified Molecular Input LineEntry System (hereinafter referred to as SMILES), SMARTS, or InChI. ECFP can digitize the molecular structure of the raw material by, for example, inputting the information of the raw material represented by SMILES into an existing library.

[0061] In addition, as another example of a method for quantifying the molecular structure of a raw material, there is a method for quantifying the structural characteristics of the molecule by extracting calculable physical property values ​​(molecular weight, radical charge number, or number of electrons, etc.) from the molecular structure of the raw material and expressing it as a vector (column: type, value: number). The physical property values ​​can be extracted, for example, by inputting information on the raw material represented by SMILES notation into an existing library.

[0062] The design support device 10 receives information input by the operator to the user terminal 12 and performs processing for supporting the design of the material. The design support device 10 transmits the processing result information to the user terminal 12 so that the user terminal 12 displays the processing result information.

[0063] Figure 1 The information processing system 1 can be implemented by a design support device 10 having a Web server function and a user terminal 12 that executes a Web application through a Web browser function. Figure 1 The information processing system 1 can be realized by processing by an application installed in the user terminal 12 in cooperation with a program installed in the design support apparatus 10 .

[0064] It should be noted that Figure 1 The information processing system 1 is only an example, and it is obvious that there are various system configuration examples depending on the use and purpose. For example, the design support device 10 can be implemented by multiple computers or as a cloud computing service. In addition, Figure 1 The information processing system 1 may be implemented by an independent computer.

[0065] <Hardware Configuration>

[0066] Figure 1 The design support device 10 and the user terminal 12 are composed of, for example, Figure 2 The present invention is implemented by a computer 500 having the hardware configuration shown.

[0067] Figure 2 This is a diagram showing a hardware configuration of an example of a computer according to the present embodiment. Figure 2 The computer 500 includes an input device 501, a display device 502, an external I / F 503, a RAM 504, a ROM 505, a CPU 506, a communication I / F 507, and a HDD 508, which are connected to each other via a bus B. It should be noted that the input device 501 and the display device 502 may be used in a connected manner.

[0068] The input device 501 is a touch panel, operation keys or buttons, keyboard or mouse, etc. for the user to input various signals. The display device 502 is composed of a display such as a liquid crystal or organic EL for displaying a screen, a speaker for outputting sound data such as voice or sound, etc. The communication I / F 507 is an interface for the computer 500 to perform data communication.

[0069] In addition, HDD508 is an example of a non-volatile storage device that stores programs or data. The stored programs or data include OS, which is basic software for controlling the entire computer 500, and application programs that provide various functions on OS. It should be noted that the computer 500 can use a drive device (such as a solid state drive: SSD, etc.) that uses flash memory as a storage medium instead of HDD508.

[0070] The external I / F 503 is an interface with an external device. The external device includes a recording medium 503a, etc. Thus, the computer 500 can read and / or write to the recording medium 503a via the external I / F 503. The recording medium 503a includes a floppy disk, a CD, a DVD, an SD memory card, a USB memory, etc.

[0071] ROM 505 is an example of a nonvolatile semiconductor memory (storage device) that can save programs or data even when the power is turned off. Programs or data such as BIOS, OS settings, and network settings that are executed when the computer 500 is started are stored in ROM 505. RAM 504 is an example of a volatile semiconductor memory (storage device) that temporarily saves programs or data.

[0072] The CPU 506 is a computing device that realizes the control and functions of the entire computer 500 by reading programs or data from a storage device such as the ROM 505 or the HDD 508 to the RAM 504 and executing processing. The computer 500 according to the present embodiment can realize various functions of the design support device 10 and the user terminal 12 described later by executing programs.

[0073] <Functional Configuration>

[0074] The functional configuration of the information processing system 1 according to the present embodiment will be described. Figure 3 is a functional structure diagram of an example of an information processing system according to this embodiment. Figure 3 In the structural diagram, parts that are not necessary for the description of this embodiment are appropriately omitted.

[0075] Figure 3 The design support device 10 of the information processing system 1 shown has a request receiving unit 20, a response sending unit 22, a registration receiving unit 24, a design condition input receiving unit 26, a required characteristic input receiving unit 28, a range information input receiving unit 30, a control unit 32, a feature quantity generating unit 34, an exhaustive search point generating unit 36, a material characteristic predicting unit 38, a design condition suggestion unit 40, a raw material information storage unit 42, a machine learning model storage unit 44, and a display control unit 46. The user terminal 12 has an information display unit 50, an operation receiving unit 52, a request sending unit 54, and a response receiving unit 56.

[0076] The information display unit 50 displays a screen for receiving information input from the operator and information on the execution result of the process of the design support device 10 on the display device 502. The operation receiving unit 52 receives the operation of the operator such as the input of information. The request sending unit 54 sends a request for a process corresponding to the information input from the operator to the design support device 10. The response receiving unit 56 receives a response to the request for the process sent by the request sending unit 54 from the design support device 10.

[0077] The request receiving unit 20 receives a request for processing from the user terminal 12. The response sending unit 22 responds with the execution result of the processing corresponding to the request for processing. The registration receiving unit 24 receives input of the name and attribute information of the raw material to be newly registered, and stores it in the raw material information storage unit 42. The raw material information storage unit 42 stores the name and attribute information of a plurality of registered raw materials.

[0078] The design condition input receiving unit 26 receives input of design condition information including information of a plurality of raw materials stored in the raw material information storage unit 42 from the operator. The required characteristic input receiving unit 28 receives input of required characteristic information of a material to be blended with the plurality of raw materials from the operator. The range information input receiving unit 30 receives input of range information of design condition information including information of a plurality of raw materials stored in the raw material information storage unit 42.

[0079] As described below, the feature quantity generating unit 34 generates a feature quantity of the raw material that can be input to the machine learning model after learning based on the information of the properties of the raw material. The exhaustive search point generating unit 36 ​​generates a predetermined number (e.g., 1,000) of exhaustive search points at random or with a predetermined step length within the range of the range information of the design condition information. The exhaustive search point is a combination of the composition ratio or the blending amount of a plurality of raw materials used for material blending. The blending amount of each raw material is selected, for example, within the range indicated by the lower limit value and the upper limit value of the blending amount.

[0080] The material property prediction unit 38 predicts the material property based on the received design condition information using the learned machine learning model. The design condition suggestion unit 40 suggests candidates for design condition information that meets the input required property information using the learned machine learning model. The design condition suggestion unit 40 predicts the material property corresponding to each exhaustive search point using the learned machine learning model and extracts the exhaustive search point that meets the property of the required property information, thereby obtaining candidates for design condition information that meets the desired required property information.

[0081] The machine learning model storage unit 44 stores a machine learning model that has learned the correspondence between the design condition information of the material and the characteristic information of the material. The design condition information of the material includes information such as the type and amount of the raw material, and the process conditions. The process conditions are, for example, the treatment temperature and treatment time of heat treatment, etc. In addition, the characteristic information of the material is viscosity, glass transition temperature, molecular weight, and acid value.

[0082] The control unit 32 controls the request receiving unit 20, the response sending unit 22, the registration receiving unit 24, the design condition input receiving unit 26, the required characteristic input receiving unit 28, the range information input receiving unit 30, the feature quantity generating unit 34, the exhaustive search point generating unit 36, the material characteristic predicting unit 38, the design condition suggesting unit 40, and the display control unit 46. The display control unit 46 controls so that the material characteristic predicted by the material characteristic predicting unit 38 or the candidate of the design condition information suggested by the design condition suggesting unit 40 is displayed on the user terminal 12.

[0083] It should be noted that Figure 3 The configuration diagram is only an example. The configuration of the information processing system 1 according to the present embodiment may be implemented by various configurations. For example, the raw material information storage unit 42 and the machine learning model storage unit 44 may have a storage device, a computer, or a cloud storage device that can communicate data with the design support device 10.

[0084] <Processing>

[0085] 《Registration of New Raw Materials》

[0086] In the information processing system 1 according to the present embodiment, the operator selects a menu for registering a new material, and thus, for example, Figure 4 and Figure 5 New material registration screen 1000 is shown.

[0087] Figure 4 and Figure 5 2 is a diagram showing an example of a new material registration screen according to the present embodiment. Figure 4 The new raw material registration screen 1000 is an example of a screen for inputting the purpose of the raw material to be newly registered, the raw material name to be added, and the SMILES of the raw material. The raw material name to be added is the name of the raw material to be newly registered, and is an example of information for the operator to identify the raw material. The purpose of the raw material and the SMILES of the raw material are examples of information on the attributes of the raw material.

[0088] For example Figure 5 As shown in FIG. 1 , the purpose of the raw material is an example of information indicating the purpose of using the raw material, such as monomer (main chain), monomer (side chain), polymerization initiator, solvent, or catalyst. Figure 5 The new raw material registration screen 1000 shows an example in which the operator selects a usage of a raw material from among a plurality of usages of raw materials.

[0089] The SMILES of a raw material is an example of molecular structure information of a raw material to be newly registered. Figure 4 and Figure 5 The new material registration screen 1000 of FIG. 1 shows an example of inputting the molecular structure information of the material to be newly registered by SMILES notation, but it can also be input by SMARTS notation or InChI notation. Figure 4 and Figure 5 In the new raw material registration screen 1000, the molecular structure information of the raw material to be newly registered can be input in any one of the file formats for expressing molecular structures, such as MOL format, SDF format, PDB format, or CIF format. Figure 4 and Figure 5 In the new raw material registration screen 1000 , input of information of the raw material to be added can be accepted by uploading the purpose of the raw material to be added, the name of the raw material to be added, and a list of SMILES of the raw material.

[0090] The operator can select a menu for displaying registered raw materials, thereby displaying, for example, Figure 6 A registered raw material list screen 1100 is shown. Figure 6 This is a diagram showing an example of the registered raw material list screen. Figure 6 The registered raw material list screen 1100 is a list showing the raw materials from Figure 4 and Figure 5 The new raw material registration screen 1000 shown is a screen example of the addition date, purpose, name, and SMILES of the raw material newly registered and stored in the raw material information storage unit 42.

[0091] It should be noted that, in the information processing system 1 according to the present embodiment, duplicate registration of a molecular structure equivalent to a raw material already stored in the raw material information storage unit 42 is prohibited as follows.

[0092] Figure 7 FIG. 1 is a flowchart showing an example of a process for prohibiting registration of a new material having an equivalent molecular structure. In step S10, the registration receiving unit 24 of the design support device 10 receives a Figure 4 and Figure 5 The new raw material registration screen 1000 shown receives input of new raw material information.

[0093] In step S12, the registration receiving unit 24 converts the information of the molecular structure contained in the information of the new raw material received by input into the canonical SMILES of the new raw material. In step S14, the registration receiving unit 24 compares the canonical SMILES of the new raw material with the canonical SMILES of the raw material registered in the raw material information storage unit 42. It should be noted that the comparison based on the canonical SMILES is only an example, and any information that can be compared with the uniqueness of the raw material can be used.

[0094] In step S16, the registration accepting unit 24 determines whether there is a raw material registered in the raw material information storage unit 42 whose canonical SMILES is consistent with the new raw material. If there is no raw material registered in the raw material information storage unit 42 whose canonical SMILES is consistent with the new raw material, the registration accepting unit 24 determines that the new raw material does not have a molecular structure equivalent to that of the raw material stored in the raw material information storage unit 42, and registers the new raw material in the raw material information storage unit 42 in step S18.

[0095] On the other hand, if there is a raw material already registered in the raw material information storage unit 42 whose canonical SMILES is consistent with the new raw material, the registration acceptance unit 24 determines that the new raw material has a molecular structure equivalent to that of the raw material already stored in the raw material information storage unit 42, and performs the processing of step S20. In step S20, the registration acceptance unit 24 refuses to register the new raw material with the raw material information storage unit 42 to prohibit double registration of the new raw material having a molecular structure equivalent to that of the raw material already stored in the raw material information storage unit 42.

[0096] Figure 8 FIG. 1 is a flowchart showing an example of a process for prohibiting registration of a new material having an equivalent molecular structure. In step S30, the registration receiving unit 24 of the design support device 10 receives a Figure 4 and Figure 5 The new raw material registration screen 1000 shown receives input of new raw material information.

[0097] In step S32, the registration receiving unit 24 converts the information of the molecular structure contained in the information of the new raw material received as input into the canonical SMILES of the new raw material. In step S34, the registration receiving unit 24 compares the canonical SMILES of the new raw material with the canonical SMILES of the raw material with the same purpose as the new raw material and registered in the raw material information storage unit 42. It should be noted that the comparison based on the canonical SMILES is only an example, and any information that can be compared with the uniqueness of the raw material may be used.

[0098] In step S36 , the registration receiving unit 24 determines whether there is a raw material registered in the raw material information storage unit 42 that has the same usage as the new raw material and whose canonical SMILES matches the new raw material.

[0099] If there is no material registered in the material information storage unit 42 that has the same usage as the new material and whose canonical SMILES matches that of the new material, the registration accepting unit 24 registers the new material in the material information storage unit 42 in step S38 .

[0100] On the other hand, if there is a raw material already registered in the raw material information storage unit 42 whose usage is the same as that of the new raw material and whose canonical SMILES is consistent with that of the new raw material, the registration acceptance unit 24 determines that the new raw material has a molecular structure equivalent to that of the raw material with the same usage already stored in the raw material information storage unit 42, and performs the processing of step S40. In step S40, the registration acceptance unit 24 refuses to register the new raw material with the raw material information storage unit 42 to prohibit double registration of the new raw material having a molecular structure equivalent to that of the raw material with the same usage already stored in the raw material information storage unit 42.

[0101] Fig. 9 FIG. 1 is a flowchart showing an example of a process for prohibiting registration of a new material having an equivalent molecular structure. In step S50, the registration receiving unit 24 of the design support device 10 receives a Figure 4 and Figure 5 The new raw material registration screen 1000 shown receives input of new raw material information.

[0102] In step S52, the registration receiving unit 24 converts the information of the molecular structure contained in the information of the new raw material received by the input into the canonical SMILES of the new raw material. In step S54, the registration receiving unit 24 compares the canonical SMILES of the new raw material with the canonical SMILES of the raw material that has been registered in the raw material information storage unit 42 and has the same purpose as the new raw material. It should be noted that the comparison based on the canonical SMILES is only an example, and any information that can be compared with the uniqueness of the raw material can be used.

[0103] In step S56 , the registration receiving unit 24 determines whether there is a raw material registered in the raw material information storage unit 42 that has the same usage as the new raw material and whose canonical SMILES matches the new raw material.

[0104] If there is no material registered in the material information storage unit 42 that has the same usage as the new material and whose canonical SMILES matches the new material, the registration receiving unit 24 proceeds to the process of step S58. In step S58, the registration receiving unit 24 compares the name of the new material with the names of the materials registered in the material information storage unit 42.

[0105] If there is no raw material registered in the raw material information storage unit 42 and having the same name as the new raw material, the registration accepting unit 24 registers the new raw material in the raw material information storage unit 42 in step S60 .

[0106] On the other hand, in step S56 , if there is a material registered in the material information storage unit 42 that has the same usage as the new material and whose canonical SMILES matches the new material, the registration acceptance unit 24 executes the process of step S62 .

[0107] In addition, in step S58, if there is a raw material registered in the raw material information storage unit 42 that matches the name of the new raw material, the registration acceptance unit 24 performs the processing of step S62. In step S62, the registration acceptance unit 24 determines that the new raw material has a molecular structure equivalent to that of a raw material for the same purpose stored in the raw material information storage unit 42, or the new raw material has the same name as that of a raw material stored in the raw material information storage unit 42, and performs the processing of step S62. In step S62, the registration acceptance unit 24 refuses to register the new raw material to the raw material information storage unit 42 to prohibit double registration of a new raw material having a molecular structure equivalent to or the same name as that of a raw material for the same purpose stored in the raw material information storage unit 42.

[0108] 《Material Design Assistance Based on Forward Problem Analysis》

[0109] In the information processing system 1 according to the present embodiment, the operator selects a menu of assistance based on forward problem analysis designed by the material, thereby executing Fig.10 The material shown is designed for an assisted process based on forward problem analysis. Fig.10 This is a flowchart showing an example of a process of assisting material design based on forward problem analysis. Fig.11A This is a configuration diagram showing an example of design condition information of a material. Fig. 11B This is a diagram showing an example of the structure of material property prediction information.

[0110] In step S100, the design condition input receiving unit 26 of the design support device 10 obtains the design condition input by the operator at the user terminal 12, for example, Fig.11A The material design condition information shown in FIG. 1 includes the type and blending amount of the raw material stored in the raw material information storage unit 42 as information. Fig.11A The ID "polymer 1" is an example of information for identifying a material. The raw material category is an example of information for indicating the purpose of the raw material. The raw material name is an example of information for indicating the name of the raw material. The compounding amount is an example of information for indicating the amount of the raw material used for the synthesis of the material.

[0111] It should be noted that Fig.11AThe raw material named "new monomer C" is an example of a raw material registered in the raw material information storage unit 42 by new raw material registration. Fig.11A Raw materials other than the raw material name “new monomer C” are included as raw materials in the design condition information of the learning data set used for learning the machine learning model stored in the machine learning model storage unit 44.

[0112] In step S102 , the control unit 32 uses the learned machine learning model stored in the machine learning model storage unit 44 to predict the characteristics of the material acquired by the design condition input receiving unit 26 in step S100 in the following manner.

[0113] The control unit 32 reads out the type and amount of raw materials to be mixed in the material from the design condition information of the material acquired by the design condition input receiving unit 26 in step S100, and provides it to the feature value generating unit 34 to request the generating unit 34 to generate the feature value of the raw material. The feature value generating unit 34 may generate a feature value of the raw material, for example. Fig.12 The characteristic quantities of the raw materials to be mixed in the material are generated as shown.

[0114] Fig.12 1 is an explanatory diagram of an example of processing for generating characteristic quantities of raw materials to be mixed in a material. The characteristic quantity generating unit 34 reads information on the molecular structure of the raw materials expressed in SMILES notation from the raw material information storage unit 42, and the information on the molecular structure of the raw materials is an example of information on the molecular structures of "raw material A" and "raw material B" to be mixed in the material.

[0115] The feature quantity generating unit 34 converts the information of the molecular structure of the raw material into numerical information by using a method for digitizing the molecular structure of the raw material, such as ECFP. Fig.12 In the example, ECFP is used to convert the molecular structure information of "raw material A" and "raw material B" to be mixed in the material into 1024-dimensional numerical information.

[0116] The feature quantity generating unit 34 generates the feature quantity of the raw material to be input into the learning completion model by accumulating the 1024-dimensional numerical information obtained by converting the molecular structure information of "raw material A" and "raw material B" to be mixed in the material using ECFP and the mixing amount of the raw material to be mixed in the material. Fig.12 The main chain information shown can generate the characteristic quantity of the main chain raw material by Σ(the amount of the main chain raw material i × the ECFP vector of the main chain raw material i). Fig.12 In it is “i=A, B”.

[0117] It should be noted that if Fig.12As shown in the figure, the main chain information and the side chain information are accumulated separately. The reason is that the same raw material may be used as an initiator and as a monomer, and if the accumulation is performed without considering the use of the raw material, the correct result cannot be obtained. In the case of a composition composed of raw materials that do not need to consider the reaction order, calculations can also be performed by accumulating all materials.

[0118] The control unit 32 provides the characteristic quantity of the raw material to be mixed in the material generated by the characteristic quantity generating unit 34 to the material characteristic prediction unit 38 to request the material characteristic prediction. The material characteristic prediction unit 38 inputs the characteristic quantity of the raw material to be mixed in the material into the learned machine learning model to predict the material characteristic. The material characteristic prediction unit 38 predicts the material characteristic by using the learned machine learning model to obtain, for example, Fig. 11B The properties of the materials shown are predicted. Fig. 11B The material property prediction information shown shows the viscosity, glass transition temperature, molecular weight, and acid value of the material as examples of the material properties.

[0119] In step S104, the display control unit 46 of the design support device 10 controls so that the information input by the operator in step S100, for example, is displayed on the user terminal 12. Fig.11A The design condition information of the material shown, and the information obtained in step S102, such as Fig. 11B Property prediction information for the material shown.

[0120] pass Fig.10 The processing can predict the properties of materials that are combined with raw materials that are not included in the learning data set used for learning the machine learning model through forward problem analysis, thereby assisting in the design of materials.

[0121] 《Aids in material design based on reverse problem analysis》

[0122] In the information processing system 1 according to the present embodiment, the operator selects a menu of assistance based on reverse problem analysis designed by the material, thereby executing, for example, Fig.13 The material shown is designed based on an assisted process of reverse problem analysis. Fig.13 1 is a flowchart showing an example of a process of assisting material design based on reverse problem analysis. Fig.14 This is a diagram showing an example of required characteristic information of a material. Fig.15 This is a diagram showing an example of the range information of the material design condition information. Fig.16A and Fig. 16B This is a diagram showing an example of information on suggested materials.

[0123] In step S200, the required characteristic input receiving unit 28 of the design support device 10 obtains the required characteristic input by the operator at the user terminal 12, for example Fig.14 Information on the required properties of the materials shown. Fig.14 The required characteristic information of the material is an example of expressing the required characteristics of the material by a lower limit value and an upper limit value. Fig.14 In the table, as examples of required properties of the material, lower limits and upper limits of viscosity, glass transition temperature, molecular weight, and acid value are shown.

[0124] In step S202, the range information input receiving unit 30 obtains the range information input by the operator at the user terminal 12, for example Fig.15 The range of design condition information of the material shown is information. Fig.15 The range information of the design condition information of the material includes the raw material category, raw material name, lower limit value of the blending amount, and upper limit value of the blending amount as information. The raw material category and raw material name indicate the type of raw material. The lower limit value of the blending amount and the upper limit value of the blending amount indicate the range of the blending amount of each raw material.

[0125] in addition, Fig.15 The range information of the design condition information of the material is an example in which the operator can select the raw materials that must be mixed in the material. The operator can select the raw materials that must be mixed in the material by checking the item "must be included". Fig.15 This is an example of selecting "monomer A" and "monomer B" as the raw materials that must be mixed. It should be noted that Fig.15 The raw material named “new monomer C” is an example of the raw material registered in the raw material information storage unit 42 by the new raw material registration.

[0126] In step S204, the control unit 32 provides the range information of the material design condition information obtained in step S202 to the exhaustive search point generation unit 36 ​​to request the generation of exhaustive search points within the range information of the material design condition information. The exhaustive search point generation unit 36 ​​generates a predetermined number (e.g., 1000) of exhaustive search points within the range information of the material design condition information provided.

[0127] For example, if it is Fig.15 If the range information of the design condition information of the material is provided, the exhaustive search point is the combination of the blending amount of the raw materials "monomer A", "monomer B", "new monomer C", "polymerization initiator A", "polymerization initiator C" and "polymerization initiator D". The blending amount of each raw material is selected from the range represented by the items "lower limit value of blending amount" and "upper limit value of blending amount".

[0128] In step S206, the design condition recommendation unit 40 uses the learned machine learning model stored in the machine learning model storage unit 44 to predict the material properties based on the multiple exhaustive search points generated in step S204, and obtains the material property prediction information corresponding to the multiple exhaustive search points.

[0129] In step S208, the design condition suggestion unit 40 extracts the material property prediction information obtained in step S206 that satisfies, for example, Fig.14 The processing of step S208 is to extract the exhaustive search points that can meet the required characteristic information of the material shown in the figure from the exhaustive search points generated randomly or with a predetermined step size within the range of the design conditions. Fig.14 A process of exhaustively searching for material properties information is shown.

[0130] In step S210, the display control unit 46 of the design support device 10 controls so that candidates for design condition information and characteristic prediction information of materials to be synthesized based on the exhaustive search points extracted in step S208 are displayed as, for example, Fig.16A and Fig. 16B The information of the suggested materials shown is displayed on the user terminal 12. For example Fig.16A As information of the suggested material, characteristic prediction information of the material to be synthesized or the like based on the exhaustive search points extracted in step S208 is shown. Fig. 16B As information of the suggested material, design condition information of the material to be synthesized or the like based on the exhaustive search points extracted in step S208 is shown.

[0131] Fig.16A and Fig. 16B The information shown in the suggested materials is an example of Fig.16A The predicted properties of the recommended materials shown in Fig. 16B The design condition information of the recommended materials shown in are displayed side by side in one row by combining them using the ID as a key.

[0132] pass Fig.13 The processing can, through reverse problem analysis, recommend design condition information for materials that meet the required properties among materials that are combined with raw materials not included in the learning data set used for learning the machine learning model, thereby assisting in the design of materials.

[0133] [Other embodiments]

[0134] Regarding the material designed by the design support device 10 according to this embodiment, for example, the material design condition information can be provided to a manufacturing device that produces the material by mixing a plurality of raw materials, so that the manufacturing device produces the material by mixing the plurality of raw materials.

[0135] As described above, the information processing system 1 according to the present embodiment can provide a design support device, a design support method, a program, and an information processing system that predict the properties of a material blended with a plurality of raw materials to support the design of the material.

[0136] Although the present embodiment is described above, it should be understood that various changes in form or details can be made without departing from the spirit and scope of the claims. Although the present invention is described above based on the embodiments, the present invention is not limited to the above embodiments and various modifications can be made within the scope described in the claims. This application is based on the basic application No. 2022-149308 filed with the Japan Patent Office on September 20, 2022 as the basis for claiming priority, and all of its contents are cited here by reference.

[0137] Explanation of symbols

[0138] 1 Information Processing System

[0139] 10 Design auxiliary devices

[0140] 12 User Terminal

[0141] 18 Communication Network

[0142] 24Registration acceptance department

[0143] 26 Design condition input receiving unit

[0144] 28 Required characteristic input receiving unit

[0145] 30 Range information input receiving unit

[0146] 32. Control Unit

[0147] 34 Feature quantity generating unit

[0148] 36 Exhaustive search point generation unit

[0149] 38Material Properties Prediction Department

[0150] 40 Design Conditions Recommendation Department

[0151] 42Raw material information storage unit

[0152] 44 Machine Learning Model Storage

[0153] 46: display control unit.

Claims

1. A design assisting device, which assists the design of a material for combining multiple raw materials by using a machine learning model, wherein the machine learning model learns the correspondence between design condition information of a material for combining multiple raw materials and characteristic information of the materials, the design assisting device comprising: A raw material information storage unit configured to store information on the names and properties of a plurality of registered raw materials; a registration receiving unit configured to receive input of the name and attribute information of the raw material to be newly registered, and store the information in the raw material information storage unit; as well as The feature quantity generating unit is configured to generate feature quantities of the raw materials that can be input into the machine learning model based on information on properties of multiple raw materials selected as the design condition information and stored in the raw material information storage unit.

2. The design assistance device according to claim 1, wherein: The registration accepting unit is configured to prohibit new registration of a raw material that matches the information on the molecular structure of the raw material already stored in the raw material information storage unit.

3. The design assistance device according to claim 1, wherein: The information on the properties of the raw material includes information on the use of the raw material. The registration accepting unit is configured to prohibit new registration of a raw material that matches the information on the molecular structure of the raw material and the purpose of the raw material that are already stored in the raw material information storage unit.

4. The design assistance device according to claim 3, wherein: The registration accepting unit is configured to prohibit new registration of a raw material that matches the name of the raw material already stored in the raw material information storage unit.

5. The design assistance device according to any one of claims 1 to 4, wherein: The registration accepting unit is configured to accept input of information on a molecular structure of a raw material in any one of SMILES notation, SMARTS notation, or InChI notation as the information on the attribute of the raw material to be newly registered.

6. The design assistance device according to any one of claims 1 to 4, wherein: The registration accepting unit is configured to accept input of information on a molecular structure of a raw material in any one of a MOL format, a SDF format, a PDB format, or a CIF format as the information on the attribute of the raw material to be newly registered.

7. The design aid device according to any one of claims 1 to 6, wherein: The feature quantity generating unit is configured to generate the feature quantity of the raw material based on information on the molecular structure of the raw material using a fingerprint method.

8. The design aid device according to any one of claims 1 to 7, wherein: The feature quantity generating section is configured to generate the feature quantity of the raw material by calculating a physical property value of a molecule from a molecular structure of the raw material.

9. The design aid device according to any one of claims 1 to 8, wherein: The attribute information inputted by the registration accepting unit includes a physical property value of the raw material.

10. The design assistance device according to any one of claims 1 to 9, comprising: a design condition input receiving unit configured to receive input of the design condition information including information of a plurality of raw materials stored in the raw material information storage unit; as well as The material property prediction unit is configured to predict the property of the material using the machine learning model based on the design condition information received as input.

11. The design assistance device according to any one of claims 1 to 9, further comprising: a required characteristic input receiving unit configured to receive input of required characteristic information of the material; a range information input accepting unit configured to accept input of range information of the design condition information including information of a plurality of raw materials stored in the raw material information storage unit; as well as The design condition suggestion unit is configured to suggest candidates for the design condition information that meets the required characteristic information based on the result obtained by predicting the characteristics of the material based on the design condition information within the range of the range information using the machine learning model.

12. The design assistance device according to claim 10 or 11, further comprising: The display control unit is configured to display the predicted characteristics of the material or the recommended candidates of the design condition information.

13. A design assistance method, wherein a computer uses a machine learning model to assist in the design of a material that is combined with multiple raw materials, wherein the machine learning model learns the correspondence between design condition information of the material that is combined with multiple raw materials and characteristic information of the material, and the design assistance method comprises: a registration accepting step of accepting input of the name and attribute information of the raw material to be newly registered, and registering it in a raw material information storage unit, wherein the raw material information storage unit stores the name and attribute information of a plurality of registered raw materials; as well as The feature quantity generating step generates the feature quantity of the raw material that can be input into the machine learning model based on the information of the properties of the plurality of raw materials selected as the design condition information and stored in the raw material information storage unit.

14. A program for causing a computer to execute a process, wherein the computer assists in the design of a material for mixing multiple raw materials using a machine learning model, wherein the machine learning model learns the correspondence between design condition information of a material for mixing multiple raw materials and characteristic information of the materials, the process comprising: A registration receiving step of receiving input of the name and attribute information of the raw material to be newly registered, and registering the input into a raw material information storage unit, wherein the raw material information storage unit stores the name and attribute information of a plurality of raw materials that have been registered; as well as The feature quantity generating step generates a feature quantity of the raw material that can be input into the machine learning model based on the information on the properties of the plurality of raw materials selected as the design condition information and stored in the raw material information storage unit.

15. An information processing system, wherein a plurality of computers use a machine learning model to assist in the design of a material for combining multiple raw materials, wherein the machine learning model learns the correspondence between design condition information of a material for combining multiple raw materials and characteristic information of the materials, the information processing system comprising: A raw material information storage unit configured to store information on the names and properties of a plurality of registered raw materials; a registration receiving unit configured to receive input of the name and attribute information of the raw material to be newly registered, and store the information in the raw material information storage unit; as well as The feature quantity generating unit is configured to generate feature quantities of the raw materials that can be input into the machine learning model based on information on properties of multiple raw materials selected as the design condition information and stored in the raw material information storage unit.

Images