Information processing system, information processing method, and program
By setting a subclass of crystalline phase information in the information processing system, the combination types of crystalline phase information in the selection data are reduced, and the problem of long creation time of neural networks is solved, performance is improved and learning time is shortened.
Patent Information
- Application Number
- CN202411185168.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2023-08-28
- Filing Date
- 2024-08-27
- Publication Date
- 2025-02-28
AI Technical Summary
In qualitative analysis using machine learning, the creation of neural networks requires learning huge teacher data and spending time, especially when the amount of crystalline phase information increases, the number of combination methods increases leaps and increases learning time further.
By setting the crystallized phase information to the first and second subcategories in the information processing system, selecting data to generate an X-ray diffraction distribution map, reducing the combination types of crystallized phase information in the selection data, thereby generating more effective teacher data.
The combination mode of the mixture is reduced through subclass settings, the performance of the neural network is improved, the learning time is shortened, and high performance is maintained when there are too many data groups or learning time is short.
Smart Images
Figure CN119534505A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to an information processing system, an information processing method, and a program. Background Art
[0002] As an example of a method for analyzing the crystalline phase contained in a powder sample (hereinafter referred to as qualitative analysis), an X-ray diffraction method can be cited. In the existing qualitative analysis of the X-ray diffraction method, the crystalline phase contained in the powder sample is identified by comparing the diffraction pattern generated from the powder sample with the diffraction pattern previously recorded in the database. However, in the existing method, the amount of data on the crystalline phase recorded in the database is large, and there is a problem that it takes a lot of time to compare the diffraction pattern with the information recorded in the database, and the correct answer rate of complex diffraction patterns is low.
[0003] In recent years, as a method to improve existing qualitative analysis, qualitative analysis using machine learning is being studied. In qualitative analysis using machine learning, a neural network is created that uses teacher data as learning data, which uses diffraction patterns as input data and crystalline phases contained in the diffraction patterns as output data. By inputting the measurement results of the sample under test into the neural network, the crystalline phase contained in the sample under test can be obtained.
[0004] Non-Patent Document 1 teaches a method for creating teacher data, describing that by creating a plurality of single-phase distribution maps based on crystal phase information stored in a database and adding the single-phase distribution maps, a diffraction pattern of a mixture can be simulated, and machine learning can be performed using the obtained diffraction pattern.
[0005] Prior art literature
[0006] Non-patent literature
[0007] Non-patent literature 1: Jan Schuetzke et al., Enhancing deep-learning training for phase identification in powder X-ray diffractograms, IUCrJ8 (2021) 408-420. Summary of the invention
[0008] Problem that the invention aims to solve
[0009] In the qualitative analysis using machine learning, can cite as a problem: the creation of a neural network needs to learn huge teacher data and spend time. For example, about 70 kinds of crystallization phase information are included in the cement database. The combination method of this crystallization phase information is huge, is about 10 billion, needs to learn the teacher data of huge amount. And it can be expected that when the crystallization phase information quantity included in the database increases, this combination method also can increase by leaps and bounds, further needs time. Therefore, seek the method that alleviates the learning time.
[0010] Solutions for solving problems
[0011] According to one embodiment of the present invention, an information processing system for generating an X-ray diffraction distribution diagram is provided. The information processing system has a data group acquisition unit, a subclass setting unit, a crystal phase selection unit, and a distribution diagram generation unit. The data group acquisition unit acquires a data group containing multiple crystal phase information. The subclass setting unit sets the crystal phase information contained in the data group to any one of the first subclass and the second subclass. The number of the first subclass is more than 1, the number of the second subclass is more than 0, and the total number of the first subclass and the second subclass is more than 2. The crystal phase selection unit selects crystal phase information from the first subclass or both the first subclass and the second subclass and sets it as selected data. When there are multiple numbers of the first subclass, at least one crystal phase information is included in the selected data for each first subclass. The distribution diagram generation unit generates an X-ray diffraction distribution diagram based on the selected data. BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Figure 1 This is a diagram showing an example of the system configuration of the information processing system.
[0013] Figure 2 This is a diagram showing an example of the hardware configuration of the information processing device.
[0014] Figure 3 This is a diagram showing an example of the functional configuration of an information processing device.
[0015] Figure 4 This is a flowchart showing an example of information processing during learning by the information processing device.
[0016] Figure 5 It is a diagram showing an example of a setting screen.
[0017] Figure 6 It is a diagram showing a modified example of the setting screen.
[0018] Figure 7 This is a flowchart showing an example of information processing during inference by the information processing device.
[0019] Figure 8This is a conceptual diagram showing an example of a process for determining a crystal phase.
[0020] Fig. 9 This is a diagram showing the difference in the learning data pattern depending on whether there is a subclass or not.
[0021] Fig.10 This is a diagram showing an example of comparison of performance depending on the presence or absence of subclasses.
[0022] Fig.11 This is a diagram showing an example of comparison of learning time depending on the presence or absence of subclasses.
[0023] Description of Reference Numerals
[0024] 100: Information processing device
[0025] 110: X-ray diffraction device
[0026] 150: Network
[0027] 210: Control Department
[0028] 220: Storage
[0029] 230: Input interface
[0030] 240: Output interface
[0031] 250: Ministry of Communications
[0032] 310: Data group acquisition unit
[0033] 320: Subclass Setting Department
[0034] 330: Crystal phase selection department
[0035] 340: Distribution map generation unit
[0036] 350: Neural Network Generation Department
[0037] 360: Measurement data acquisition unit
[0038] 370: Measurement Data Analysis Department
[0039] 500: Setting screen
[0040] 510: Subclass settings window
[0041] 511: Crystal phase information display column
[0042] 512: Subcategory settings column
[0043] 520: Subclass type setting window
[0044] 521: Subcategory display column
[0045] 522: Type setting column
[0046] 530: Register button
[0047] 600: Setting screen
[0048] 610: Subclass settings window
[0049] 611: Crystal phase information display column
[0050] 612: Subcategory settings column
[0051] 630: Register button
[0052] 1000: Information processing system. DETAILED DESCRIPTION
[0053] Hereinafter, embodiments of the present invention will be described using the drawings. Various features described in the embodiments described below can be combined with each other.
[0054] In this specification, "unit" may include, for example, a combination of hardware resources implemented by a circuit in a broad sense and information processing of software that can be specifically implemented by these hardware resources. In addition, in Embodiment 1, various information is involved, and this information can be represented by the high and low values of the signal as a bit set of a binary number composed of 0 or 1, and can perform communication and calculation on a circuit in a broad sense.
[0055] In addition, a circuit in a broad sense refers to a circuit realized by at least appropriately combining a circuit, a circuit type, a processor, and a memory, etc. That is, it includes an application-specific integrated circuit (ASIC), a programmable logic device (for example, a simple programmable logic device (SPLD), a complex programmable logic device (CPLD), and a field programmable gate array (FPGA)).
[0056] <Implementation method 1>
[0057] 1. System structure diagram
[0058] Figure 1 1 is a diagram showing an example of the system configuration of the information processing system 1000. Figure 1As shown, the information processing system 1000 includes an information processing device 100 and an X-ray diffraction device 110 as a system configuration. The information processing device 100 and the X-ray diffraction device 110 are connected to each other via a network 150 so as to be communicable. The information processing device 100 is a device that generates an X-ray diffraction profile and performs the processing of the first embodiment including the modified examples described later. The X-ray diffraction device 110 is a device that performs measurement by analyzing the diffraction caused by the result of scattering and interference of the X-rays by the electrons around the atoms of the sample when the X-rays are irradiated to the sample.
[0059] Here, the information processing system described in the claims may be composed of a plurality of devices or a single device. When the information processing system described in the claims is composed of a single device, an example of the device is the information processing device 100. When the information processing system described in the claims is composed of a plurality of devices, examples of the plurality of devices are the information processing device 100 and the X-ray diffraction device 110, or a cloud system that provides the functions of the information processing device 100, etc.
[0060] 2. Hardware composition
[0061] Figure 2 1 is a diagram showing an example of the hardware configuration of the information processing device 100. Figure 2 As shown, the information processing device 100 includes a control unit 210, a storage unit 220, an input interface (hereinafter referred to as input I / F) 230, an output interface (hereinafter referred to as output I / F) 240, and a communication unit 250 as a hardware configuration.
[0062] The control unit 210 is a CPU (Central Processing Unit) and the like, and controls the entire information processing device 100. The control unit 210 implements the following functions by executing processing based on a program stored in the storage unit 220. Figure 3 The functional configuration of the information processing device 100 shown in FIG. Figure 4 and Figure 7 Processing of the flowchart shown, etc.
[0063] The storage unit 220 is any one of a HDD (Hard Disk Drive), a ROM (Read Only Memory), a RAM (Random Access Memory), and an SSD (Solid State Drive), or any combination thereof, and stores programs and data used by the control unit 210 when executing processing based on the programs. The storage unit 220 is an example of a storage medium. In addition, in this specification, it is described that the data used by the control unit 210 when executing processing based on the programs is stored in the storage unit 220, but it can also be stored in the storage unit of other devices that can communicate with the information processing device 100. That is, as long as the control unit 210 can refer to it, the data can be stored in the storage unit of any device.
[0064] The input I / F 230 is an interface for connecting an input device to the information processing device 100 . Examples of the input device include a keyboard and a mouse. Information input from the input device is transmitted to the control unit 210 via the input I / F 230 .
[0065] The output I / F 240 is an interface for connecting an output device to the information processing device 100. Examples of the output device include a display, etc. Information output via the output I / F 240 based on the control of the control unit 210 is output to the output device.
[0066] The communication unit 250 is a NIC (Network Interface Card) or the like, connects the information processing device 100 to the network 150, and is responsible for communication with other devices.
[0067] 3. Functional composition
[0068] Figure 3 1 is a diagram showing an example of the functional configuration of the information processing device 100. Figure 3 As shown, the information processing device 100 includes a data group acquisition unit 310, a subclass setting unit 320, a crystal phase selection unit 330, a distribution map generation unit 340, a neural network generation unit 350, a measurement data acquisition unit 360 and a measurement data analysis unit 370 as functional components.
[0069] The data group acquisition unit 310 acquires a data group containing multiple crystal phase information from the storage unit 220, etc. Here, the crystal phase information refers to the information that can calculate the diffraction pattern, for example, the crystal structure information, the intensity information at a certain angle (2θ), etc. can be cited. The data group is preferably a collection of structured information such as a database, but it may not be such a data group. The data group is not limited to the data group provided, and it may also be that the user collects multiple crystal phase information and collects it as a data group.
[0070] The subclass setting unit 320 sets the crystal phase information contained in the data group to any one of the first and second subclasses. Here, the number of the first subclass is greater than 1, the number of the second subclass is greater than 0, and the total number of the first and second subclasses is greater than 2. Here, the crystal phase information contained in the data group must be classified as any one of the first and second subclasses. Assuming that there is crystal phase information that is not classified as any subclass, it is considered to be classified as the second subclass. The subclass setting unit 320 can be set based on the user's input, or the subclass setting unit 320 can be set automatically. In addition, it is preferred that the subclass is determined based on the type of analysis object or the purpose of analysis. As the type of analysis object, cement, battery materials, etc. can be cited, and as the purpose of analysis, the analysis of impurity components, the determination of polymorphs, etc. can be cited.
[0071] Crystallization phase selection unit 330 selects crystallization phase information from the 1st subclass or the 1st subclass and the 2nd subclass and is set as selected data. Here, when the quantity of the 1st subclass exists multiple, at least 1 crystallization phase information is included by each 1st subclass in the selected data. By selecting crystallization phase information like this, the kind of combination of crystallization phase information included in the selected data can be reduced. For example, by setting the crystallization phase information considered as the main component to the 1st subclass, the crystallization phase information considered as impurity to the 2nd subclass, the selected data that must include the main component can be created. The above-mentioned record does not hinder the crystallization phase information considered as impurity to be set to the 1st subclass, and it can also be set to the 1st subclass when it is known that impurities must be included. In addition, no matter how to set the subclass, by setting more than 1 in the 1st subclass, the kind of combination of crystallization phase information included in the selected data can be reduced, and the effect of embodiment 1 can be enjoyed.
[0072] The distribution map generation unit 340 generates an X-ray diffraction distribution map based on the selection data selected by the crystal phase selection unit 330. In more detail, the distribution map generation unit 340 preferably generates a plurality of single-phase distribution maps from the selection data, and combines the generated plurality of single-phase distribution maps to generate an X-ray diffraction distribution map. Here, the combination of distribution maps refers to adding the intensity information using a plurality of crystal phase information and their mixing ratios, and the generated X-ray diffraction distribution map becomes the X-ray diffraction distribution map of the mixture. In addition, the above description shows a method for generating an X-ray diffraction distribution map of a mixture with good efficiency, and does not prevent the X-ray diffraction distribution map of the mixture from being directly generated from the selection data.
[0073] The neural network generator 350 generates a neural network that performs machine learning on the teacher data using the X-ray diffraction profile generated by the profile generator 340 as input data and the selected data as output data. The neural network that has completed the learning after machine learning (hereinafter also referred to as a neural network) is stored in the storage unit 220 or the like.
[0074] The measurement data acquisition unit 360 acquires an X-ray diffraction distribution diagram as measurement data. The measurement data is generated by the X-ray diffraction device 110, sent to the information processing device 100, and stored in the storage unit 220 or the like. The measurement data acquisition unit 360 acquires the measurement data from the storage unit 220 or the like.
[0075] The measurement data analysis unit 370 can output the inference result of inferring the crystal phase information contained in the measurement data by inputting the measurement data into the learned neural network. The measurement data analysis unit 370 can output the inference result by displaying the inference result on the output device via the output I / F 240, etc., or can output the inference result by writing the inference result to a file, etc. and storing it in the storage unit 220. Here, the inference result refers to the result of inferring the crystal phase information contained in the measurement data, which is a continuous value from 0 to 1 related to the probability of containing a certain crystal phase.
[0076] The measurement data analysis unit 370 can also obtain the inference result of inferring the crystal phase information contained in the measurement data by inputting the measurement data into the neural network, and output the crystal phase information contained in the measurement data based on the inference result. As a determination method for the crystal phase information contained in the measurement data, for example, a certain threshold value can be used to determine the inference result. The measurement data analysis unit 370 outputs at least one of the inference result or the crystal phase information contained in the measurement data.
[0077] 4. Information Processing
[0078] (1) When studying
[0079] Figure 4 1 is a flowchart showing an example of information processing during learning by the information processing device 100 .
[0080] In step S401 , the data group acquisition unit 310 acquires a data group including a plurality of crystal phase information from the storage unit 220 or the like.
[0081] In step S402 , the subclass setting unit 320 controls the output device to display a setting screen for setting the plurality of crystal phase information included in the data set to either the first subclass or the second subclass based on the plurality of crystal phase information included in the data set.
[0082] Figure 55 is a diagram showing an example of a setting screen 500. The setting screen 500 is composed of a subclass setting window 510, a subclass type setting window 520, and a registration button 530. In addition, the subclass setting window 510 is composed of a crystal phase information display column 511 and a subclass setting column 512, and the subclass type setting window 520 is composed of a subclass display column 521 and a subclass type setting column 522. As described above, in the subclass type setting column 522, any one of the first subclass and the second subclass can be set, and which subclass of the first subclass and the second subclass each subclass set in the subclass setting column 512 belongs to is set. Setting the crystal phase information contained in the data group to the first subclass and the second subclass means setting both the subclass and the subclass type for the crystal phase information.
[0083] The crystal phase information included in the data set is displayed in the crystal phase information display column 511. For each crystal phase, the user can enter the subclass name in the subclass setting column 512, or select it from the list box. In addition, the user can also automatically assign subclasses based on the information included in the data set without entering any information.
[0084] The subcategory set in the subcategory setting column 512 is displayed in the subcategory display column 521. In the subcategory type setting column 522, for each subcategory displayed in the subcategory display column 521, either the first subcategory or the second subcategory is selected and set. In this case, too, it is possible to select from a list box or to select which of the first subcategory and the second subcategory is selected by a check box. Figure 5 In the example, the subcategory name setting window 510 and the subcategory type setting window 520 are arranged on the same screen, but they may be present on different screens. By pressing the registration button 530, the information related to the subcategory setting is updated.
[0085] In addition, the list box is an example of a GUI component, and can be any GUI component as long as it can set a subclass for the crystal phase information. Figure 5 In the example, the subcategory setting column 512 and the subcategory type setting column 522 are separated, but it is also possible to set both the subcategory and the subcategory type at once as in the n-th first subcategory.
[0086] Figure 6 5 is a diagram showing a modified example of the setting screen 500. Figure 5 The setting screen 600 is composed of a subclass setting window 610 and a registration button 630, and the subclass setting window 610 is composed of a crystal phase information display column 611 and a subclass setting column 612. In the subclass setting column 612, the subclass name can also be recorded only for the crystal phase belonging to the first subclass. In this case, the crystal phase without a subclass name can be regarded as the second subclass.
[0087] The subclass setting unit 320 sets the crystal phase information included in the data set to any one of the first subclass and the second subclass based on the operation via the setting screen. Figure 6 As shown, when it does not belong to any subclass, it is regarded as the second subclass, and the subclass division is performed in a manner that does not include the crystal phase information that does not belong to any subclass.
[0088] In step S403, the crystal phase selection unit 330 selects crystal phase information from the first subclass or both the first subclass and the second subclass and sets the selected data. Here, when there are multiple first subclasses, the selected data includes at least one crystal phase information for each first subclass.
[0089] In step S404 , the profile generating section 340 generates an X-ray diffraction profile (mixture profile) based on the selection data selected by the crystal phase selecting section 330 .
[0090] In step S405 , the neural network generation unit 350 generates a neural network that performs machine learning on teacher data having the X-ray diffraction profile generated by the profile generation unit 340 as input data and having the selected data as output data.
[0091] (2) Inference
[0092] Figure 7 1 is a flowchart showing an example of information processing during inference by the information processing device 100 .
[0093] In step S701, the measurement data acquisition unit 360 acquires measurement data. The measurement data acquisition unit 360 acquires measurement data measured by the X-ray diffraction device 110 from the storage unit 220 or the like.
[0094] In step S702, the measurement data analysis unit 370 inputs the measurement data into the learned neural network, and obtains the inference result of inferring the crystal phase information included in the measurement data from the learned neural network. Alternatively, the crystal phase information included in the measurement data may be obtained based on the inference result.
[0095] Figure 8 1 is a conceptual diagram showing an example of the process of obtaining crystal phase information. The probability of each crystal phase being included in the measurement data is output as a continuous value of 0-1. Figure 8 The probability is recorded in . A threshold is set for the inference result, and the crystal phase information included in the measurement data is obtained.
[0096] In step S703, the measurement data analysis unit 370 outputs at least one of the inference result or the crystal phase information calculated based on the inference result. The measurement data analysis unit 370 can output the inference result by displaying the inference result on an output device via the output I / F 240, or can output the inference result by writing the inference result to a file, etc. and storing it in the storage unit 220.
[0097] 5. Implementation Method
[0098] In the embodiment, the description is given assuming that the number of pieces of crystal phase information included in the first subclass and the second subclass is 2 or more. In addition, in the embodiment, the description is given taking the case where the type of the analysis target is cement as an example.
[0099] When the type of the analysis object is cement, the subclass setting unit 320 may set the subclass as the first subclass in a manner including C3S (tricalcium silicate; Alite: tricalcium silicate), C2S (dicalcium silicate; Belite: dicalcium silicate), C3A (tricalcium aluminate; Aluminate: aluminate) and C4AF (tetracalcium aluminoferrite; Ferrite: ferrite).
[0100] In the manufacturing process of cement, C3S, C2S, C3A, and C4AF are used as raw materials. C3S, C2S, C3A, and C4AF each have multiple crystal phases, but at least one crystal phase information is selected from each of C3S, C2S, C3A, and C4AF. Therefore, it can be considered that by setting C3S, C2S, C3A, and C4AF as the first subclass and setting the impurity phase as the second subclass, efficient teacher data can be created.
[0101] The crystal phase information included in C3S, C2S, C3A, and C4AF includes, for example, the following information.
[0102] ·C3S
[0103] The chemical formula is the same: Ca3(SiO4)O, but there are different crystal systems, such as monoclinic, rhombohedral, and hexagonal.
[0104] C2S
[0105] The chemical formula is the same as Ca2(SiO4), but there are different crystal systems, including monoclinic, orthorhombic, tetragonal, rhombohedral, and hexagonal.
[0106] ·C3A
[0107] There are chemical formulas based on Ca3(Al2O6) with Na, Fe or Si added, and chemical formulas with different composition ratios. Different crystal systems include orthorhombic system, cubic system, and the like.
[0108] ·C4AF
[0109] The chemical formula includes Ca2FeAlO5 as the base with Ca added. The crystal system includes orthorhombic system and the like.
[0110] Fig. 9 The first figure shows the difference in the pattern of the combination of mixtures with or without subclasses. The following figure shows the verification results of the effect of setting subclasses on the data set of cement containing 72 and 124 crystalline phases. Generally, cement is composed of 4 main components (C3S, C2S, C3A, C4AF) and trace components. Each main component has multiple crystalline phases. Fig. 9 In the example of , the information processing device 100 sets a total of five subcategories for each main component and trace component. Fig. 9 , the difference in the pattern of the combination of mixtures (8 phases) with or without subclasses is shown. Thus, as the number of data sets increases, the pattern of the combination of mixtures increases dramatically.
[0111] Fig.10 1 is a diagram showing an example of performance comparison according to the presence or absence of subclasses. In both the case of using subclasses and the case of not using subclasses, the same amount of teacher data was created and the neural network was learned. Then, 1,000 low-strength cement simulation distribution maps were input to the learned neural network, and the crystal phase information contained in the simulation distribution maps was inferred.
[0112] Here, the F1 value is defined by the following formula.
[0113]
Mathematical formula 1
[0114] Recall=TP / (TP+FN)
[0115] Precision=TP / (TP+FP)
[0116]
[0117] TP, FP, and FN represent true positive, false positive, and false negative, respectively. The closer the F1 value is to 1, the less FP and FN are well balanced.
[0118] When the number of data sets is 72 and when the number of data sets is 124, the score is higher when there are subclasses than when there are no subclasses. In addition, the following results are obtained: when there are no subclasses, the F1 value decreases significantly as the number of data sets increases, while when there are subclasses, the F1 value remains roughly constant even if the number of data sets increases. Regarding this, the mode of reducing the combination of mixtures by using subclasses is considered to be the main reason, indicating that it is not easily affected by the increase in the number of data sets.
[0119] Fig.11 1 is a diagram showing an example of comparison of learning time with and without subclasses. In both cases of using subclasses and not using subclasses, the same amount of teacher data was created and the neural network was learned. Then, 1,000 high-strength cement simulation distribution maps were input to the learned neural network, and the crystal phase information contained in the simulation distribution maps was inferred. Fig.11 The figure shows the evolution of the F1 value during the learning process. The case with subclasses achieves a high score at an earlier stage than the case without subclasses. This means that high performance can be achieved in a short time. In addition, Fig.11 The epoch in refers to the demarcation unit of learning. Fig.11 In the example above, 500,000 data points are learned in each round. The number of rounds is proportional to the learning time.
[0120] According to the process of the embodiment, by setting subclasses for target data groups, it is possible to restrict the pattern of combinations of mixtures in the learning data.
[0121] As a result, the performance of the neural network can be improved compared to the case where no subclass is set. In addition, it is shown that setting a subclass is useful when the number of data sets is large or when learning is performed in a short learning time.
[0122] (Variant 1)
[0123] Modifications of the embodiment will be described.
[0124] In the embodiment, cement is used as an example of the type of the analysis object. However, the analysis object is not limited to cement. As another example of the analysis object, there is a battery material. In the modification 1, when the type of the analysis object is a battery material, the subclass setting unit 320 sets the subclass as the first subclass in a manner that includes at least one of a negative electrode material, a positive electrode material, a solid electrolyte, and a current collector.
[0125] Batteries are made by combining multiple materials such as positive electrode materials, negative electrode materials, solid electrolytes, and current collectors. In addition, various types of materials are used for positive electrode materials, negative electrode materials, solid electrolytes, and current collectors. Therefore, by setting positive electrode materials, negative electrode materials, solid electrolytes, and current collectors as the first subclass, teacher data can be created efficiently.
[0126] The crystal phase information contained in the positive electrode material, the negative electrode material, the solid electrolyte, and the current collector includes, for example, the following information.
[0127] Cathode materials
[0128] Layered rock salt type LiCoO2 (LCO), LiNiCoAlO2 (NCA), spinel type LiMn2O4 (LMO), olivine type LiFePO4 (LFP), sulfur-based sulfur
[0129] Negative electrode materials
[0130] Carbon-based graphite, metal-based Sn or Si, oxide-based SiO, metal-based Li
[0131] Solid electrolyte
[0132] Sulfide-based solid electrolytes, oxide-based solid electrolytes
[0133] ·Current Collector
[0134] Cu foil, Al foil
[0135] Furthermore, a part of the positive electrode material and the negative electrode material may contain a binder such as polyvinylidene fluoride (PVDF) or styrene butadiene rubber (SBR).
[0136] According to Modification 1, battery materials can also be used as the analysis object. In addition, according to Modification 1, the same effects as those of the embodiment can be achieved.
[0137] (Variant 2)
[0138] In the embodiment and modification example 1, examples are shown in which the number of crystal phase information included in the first subclass and the second subclass is 2 or more respectively, but in modification example 2, an example is shown in which the number of crystal phase information included in the first subclass is 1.
[0139] In chemical synthesis, synthesis of pharmaceuticals, etc., the target crystalline phase is sometimes clear. In this case, by setting the target crystalline phase as the first subclass and setting other compounds as the second subclass, it is possible to identify impurity components contained outside the target crystalline phase.
[0140] Furthermore, it can also be provided in each of the aspects described below.
[0141] (1) An information processing system is an information processing system for generating an X-ray diffraction distribution map, comprising a data group acquisition unit, a subclass setting unit, a crystal phase selection unit, and a distribution map generation unit, wherein the data group acquisition unit acquires a data group containing a plurality of crystal phase information, the subclass setting unit sets the crystal phase information contained in the data group to any one of a first subclass and a second subclass, the number of the first subclass is greater than 1, the number of the second subclass is greater than 0, and the total number of the first subclass and the second subclass is greater than 2, the crystal phase selection unit selects crystal phase information from the first subclass, or both the first subclass and the second subclass, and sets it as selected data, and when there are multiple numbers of the first subclass, at least one crystal phase information is included in the selected data for each of the first subclasses, and the distribution map generation unit generates an X-ray diffraction distribution map based on the selected data.
[0142] (2) In the information processing system described in (1) above, the number of pieces of crystal phase information included in the first subclass and the second subclass is respectively 2 or more.
[0143] (3) In the information processing system described in (1) or (2) above, the type of analysis object is cement, including C3S (tricalcium silicate; Alite: tricalcium silicate), C2S (dicalcium silicate; Belite: dicalcium silicate), C3A (tricalcium aluminate; Aluminate: aluminate) and C4AF (tetracalcium aluminoferrite; Ferrite: ferrite) as the first subclass.
[0144] (4) In the information processing system described in (1) or (2) above, the type of analysis object is battery material, including at least one of negative electrode material, positive electrode material, solid electrolyte, and collector as the first subclass.
[0145] (5) In the information processing system described in any one of (1) to (4) above, the subcategory setting unit displays a setting screen, and sets the crystal phase information contained in the data group to any one of the first subcategory and the second subcategory based on the operation via the setting screen.
[0146] (6) In the information processing system described in any one of (1) to (5) above, the profile generator generates a plurality of single-phase profiles from the selected data, and generates the X-ray diffraction profile by combining the generated plurality of single-phase profiles.
[0147] (7) In the information processing system described in any one of (1) to (6) above, there is also a neural network generation unit, which generates a neural network that performs machine learning on teacher data with the generated X-ray diffraction distribution pattern as input data and the selected data as output data.
[0148] (8) The information processing system described in (7) above also includes a measurement data acquisition unit and a measurement data analysis unit, wherein the measurement data acquisition unit acquires an X-ray diffraction distribution diagram as measurement data, and the measurement data analysis unit outputs an inference result of inferring the crystal phase information contained in the measurement data by inputting the measurement data into the neural network.
[0149] (9) The information processing system described in (7) above also includes a measurement data acquisition unit and a measurement data analysis unit, wherein the measurement data acquisition unit acquires an X-ray diffraction distribution diagram as measurement data, and the measurement data analysis unit obtains an inference result of inferring the crystal phase information contained in the measurement data by inputting the measurement data into the neural network, and outputs the crystal phase information contained in the measurement data based on the inference result.
[0150] (10) An information processing method is an information processing method executed by an information processing system for generating an X-ray diffraction distribution diagram, the method comprising obtaining a data group including a plurality of crystal phase information, setting the crystal phase information included in the data group to any one of the first subclass and the second subclass, wherein the number of the first subclass is greater than 1, the number of the second subclass is greater than 0, and the total number of the first subclass and the second subclass is greater than 2, selecting crystal phase information from the first subclass or both the first subclass and the second subclass and setting it as selected data, and in the case where there are multiple numbers of the first subclass, including at least one crystal phase information for each of the first subclasses in the selected data, and generating an X-ray diffraction distribution diagram based on the selected data.
[0151] (11) A program for causing a computer to function as the information processing system described in any one of (1) to (9) above.
[0152] Of course, it is not limited to this.
[0153] Finally, although various embodiments of the present invention have been described, these are given as examples and are not intended to limit the scope of the invention. New embodiments can be implemented in various other ways, and various omissions, substitutions, and changes can be made without departing from the scope of the subject matter of the invention. The embodiments and their variations are included in the scope and subject matter of the invention, and are included in the invention described in the claims and the scope of their equivalents.
Claims
1. An information processing system for generating an X-ray diffraction distribution diagram, characterized in that: It has a data set acquisition unit, a subclass setting unit, a crystal phase selection unit and a distribution map generation unit, The data group acquisition unit acquires a data group including a plurality of crystal phase information. The subclass setting unit sets the crystal phase information included in the data group as any one of the first subclass and the second subclass, The number of the first subclass is 1 or more, the number of the second subclass is 0 or more, and the total number of the first subclass and the second subclass is 2 or more, The crystal phase selection unit selects crystal phase information from the first subclass, or both the first subclass and the second subclass, and sets the selected data. When there are a plurality of first subclasses, the selected data includes at least one crystal phase information for each first subclass. The profile generating unit generates an X-ray diffraction profile based on the selection data.
2. The information processing system according to claim 1, wherein: The number of crystal phase information included in the first subclass and the second subclass is respectively greater than or equal to 2.
3. The information processing system according to claim 1, wherein: The type of analysis object is cement. The first subclass includes C3S, i.e., tricalcium silicate, C2S, i.e., dicalcium silicate, C3A, i.e., tricalcium aluminate, and C4AF, i.e., tetracalcium aluminoferrite.
4. The information processing system according to claim 1, wherein: The type of analysis object is battery materials. The first subclass includes at least one of a negative electrode material, a positive electrode material, a solid electrolyte, and a current collector.
5. The information processing system according to claim 1, wherein: The subclass setting unit The setting screen is displayed. The crystal phase information included in the data group is set to any one of the first subclass and the second subclass based on the operation via the setting screen.
6. The information processing system according to claim 1, wherein: The profile generator generates a plurality of single-phase profiles from the selected data, and generates the X-ray diffraction profile by combining the generated plurality of single-phase profiles.
7. The information processing system according to claim 1, wherein: It also has a neural network generation unit, The neural network generation unit generates a neural network that performs machine learning on teacher data using the generated X-ray diffraction profile as input data and the selected data as output data.
8. The information processing system according to claim 7, wherein: It also includes a measurement data acquisition unit and a measurement data analysis unit. The measurement data acquisition unit acquires an X-ray diffraction distribution diagram as measurement data, The measurement data analysis unit outputs an inference result of inferring crystal phase information included in the measurement data by inputting the measurement data into the neural network.
9. The information processing system according to claim 7, wherein: It also includes a measurement data acquisition unit and a measurement data analysis unit. The measurement data acquisition unit acquires an X-ray diffraction distribution diagram as measurement data, The measurement data analysis unit obtains an inference result of inferring the crystal phase information included in the measurement data by inputting the measurement data into the neural network, and outputs the crystal phase information included in the measurement data based on the inference result.
10. An information processing method, which is an information processing method executed by an information processing system for generating an X-ray diffraction distribution diagram, characterized in that: Obtain a data set containing information on multiple crystal phases, The crystal phase information contained in the data set is set to any one of the first subclass and the second subclass, The number of the first subclass is 1 or more, the number of the second subclass is 0 or more, and the total number of the first subclass and the second subclass is 2 or more, Selecting crystal phase information from the first subclass, or both the first subclass and the second subclass and setting it as selected data, When there are a plurality of first subclasses, the selected data includes at least one crystal phase information for each first subclass. An X-ray diffraction profile is generated based on the selected data.
11. A program, characterized in that It is used to make a computer function as the information processing system as described in any one of Claims 1 to 9.