Porous body design methods and porous body manufacturing methods

Abstract

This porous body design method using a computer causes the computer to execute the following processes, each multiple times: a structure generation process for virtually generating, on the computer, the three-dimensional structure of a porous body on the basis of a generation parameter value for generating a porous body; a characteristic prediction / calculation process for predicting or calculating the characteristics of the porous body having the three-dimensional structure generated by the structure generation process; an evaluation process for evaluating the characteristics of the porous body predicted or calculated by the characteristic prediction / calculation process; and an optimization process for searching for the optimal generation parameter value by changing the generation parameter value. In the porous body design method, the three-dimensional structure of the porous body is determined on the basis of a result of evaluating the characteristics of the porous body by the evaluation process.

Description

Technical Field

[0001] The present invention relates to a method for designing and manufacturing a porous body.

Background Art

[0002] Conventionally, porous bodies in which a large number of pores are three-dimensionally arranged in a material have been widely used in various applications such as purification of exhaust gas discharged from internal combustion engines such as gasoline engines and diesel engines. In the design of such porous bodies, it is necessary to determine the three-dimensional structure of the porous body so that appropriate characteristics according to the application can be obtained.

[0003] As a conventional method for designing a porous body, for example, several prototypes of the porous body are created based on the experience and intuition of a designer, and the three-dimensional structure and characteristics of these prototypes are measured respectively. Then, after grasping the relationship between the three-dimensional structure and characteristics of the porous body, the three-dimensional structure of the final porous body is determined. However, in such a design method, since it takes a lot of time and effort to create prototypes and measure their characteristics, the range in which the relationship between the three-dimensional structure and characteristics can be explored is limited. Therefore, there is a problem that the three-dimensional structure of the porous body that can obtain the desired characteristics cannot always be determined.

[0004] On the other hand, in recent years, for example, as described in Patent Document 1, a method using simulation software is also known. In such simulation software, a three-dimensional model of a porous body can be created, and the characteristics of the porous body can be obtained by calculation from this three-dimensional model.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0006] The method described in Patent Document 1 involves creating a sample from a partition section cut from an exhaust gas purification filter, and then importing the resulting continuous tomographic images obtained by X-ray CT scanning into simulation software to create a 3D model, thereby deriving the number of interconnected pores in the porous exhaust gas purification filter. Therefore, it is necessary to create the exhaust gas purification filter in advance, and the time and effort required for prototyping remains a challenge.

[0007] In view of the above-mentioned problems, the present invention aims to provide a useful technology that enables the early determination of the optimal structure in the design of porous materials. [Means for solving the problem]

[0008] The porous material design method according to the present invention is a computer-aided design method for a porous material, comprising: a structure generation process that virtually generates the three-dimensional structure of the porous material on the computer based on generation parameter values ​​for generating the porous material; An input parameter calculation process that selects at least one of the structural descriptors representing the three-dimensional structure as an input parameter for a learning model of the three-dimensional structure and calculates the value of each selected input parameter; an AI calculation process that inputs the values ​​of each input parameter calculated by the input parameter calculation process into the learning model and predicts characteristic values ​​for a predetermined characteristic among a plurality of characteristics of the porous material for the three-dimensional structure; a physical simulation process that calculates the characteristic values ​​by physical simulation based on the three-dimensional structure; an update process that updates the learning model based on the relationship between the values ​​of each input parameter obtained by the input parameter calculation process and the characteristic values ​​obtained by the physical simulation process; and the characteristic values ​​predicted by the AI ​​calculation process or calculated by the physical simulation process. An evaluation process to evaluate, Based on the evaluation results of the characteristic values ​​obtained through the evaluation process, The computer is made to perform an optimization process multiple times, which involves changing the generation parameter value to search for the optimal generation parameter value, and the evaluation process is performed as described above. Attribute values Based on the evaluation results, the three-dimensional structure of the porous body is determined. The computer executes the physical simulation process and the update process each time the AI ​​calculation process is executed a predetermined number of times. . The porous body manufacturing method according to the present invention manufactures the porous body using three-dimensional shape data based on the three-dimensional structure of the porous body determined by the porous body design method. [Effects of the Invention]

[0009] According to the present invention, a useful technology is available that enables the determination of the optimal structure early in the design of porous materials. [Brief explanation of the drawing]

[0010] [Figure 1] A block diagram showing the configuration of a porous material design apparatus according to one embodiment of the present invention. [Figure 2]A flowchart showing the processing flow of a porous material design apparatus according to one embodiment of the present invention. [Figure 3] A flowchart showing the details of the initial training model generation process. [Figure 4] A diagram illustrating a comparison between the three-dimensional structure of a porous material obtained by the porous material design apparatus of this embodiment and the three-dimensional structure of a porous material according to a comparative example. [Figure 5] A flowchart illustrating the manufacturing process of porous materials using a 3D printer. [Modes for carrying out the invention]

[0011] Embodiments of the present invention will be described below with reference to the drawings. In the following embodiment, a porous material design apparatus will be described that can be operated by a designer of porous materials to determine the three-dimensional structure of a porous material having desired properties in a relatively short time.

[0012] Figure 1 is a block diagram showing the configuration of a porous material design apparatus according to one embodiment of the present invention. The porous material design apparatus 100 shown in Figure 1 is a computer equipped with a control unit 1, a storage unit 2, a memory 3, an operation input device 4, and a display device 5, and each of these devices is connected to one another via a bus 6.

[0013] The control unit 1 is configured, for example, using a CPU (Central Processing Unit), and performs various processes and calculations to operate the porous material design apparatus 100. The control unit 1 executes programs stored in the memory unit 2 to realize the functional blocks of the structure generation unit 11, model creation unit 12, physical simulation unit 13, AI calculation unit 14, characteristic evaluation unit 15, gene evolution unit 16, model update unit 17, and data conversion unit 18. Details of these functional blocks will be described later. Note that some or all of the functions of the control unit 1 may be realized using devices other than the CPU, such as a GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), ASIC (Application Specific Integrated Circuit), etc.

[0014] The memory unit 2 is configured using, for example, a magnetic storage device such as an HDD (Hard Disk Drive) or a high-capacity, non-volatile storage device such as an SSD (Solid State Drive), and stores the program executed by the control unit 1 and various information used in the processing of the control unit 1. The information stored in the memory unit 2 includes generated parameter data 21, porous structure data 22, learning model data 23, input parameter data 24, characteristic data 25, characteristic evaluation data 26, and 3D shape data 27. Details of this data will be described later.

[0015] Memory 3 is configured using a high-speed, volatile storage device such as DRAM (Dynamic Random Access Memory), and is used as a workspace when the control unit 1 executes a program.

[0016] The operation input device 4 is a device for detecting a user's operation input, and is configured using, for example, a keyboard or a mouse. The display device 5 is a device that presents the processing result of the porous body design device 100 on a screen for the user, and is configured using, for example, a liquid crystal display, an organic EL display, or the like. Note that other computers, smartphones, etc. that can communicate with the porous body design device 100 may be used as the operation input device 4 or the display device 5.

[0017] Note that the configuration of the porous body design device 100 shown in FIG. 1 may be physically constructed on one computer or may be distributed and constructed on a plurality of computers. Further, the porous body design device 100 may be realized by a cloud computer installed on the cloud, a virtual machine operating in a virtual environment, or the like.

[0018] Next, each functional block of the structure generation unit 11, the model creation unit 12, the physical simulation unit 13, the AI calculation unit 14, the characteristic evaluation unit 15, the genetic evolution unit 16, the model update unit 17, and the data conversion unit 18 in the control unit 1 will be described. In the porous body design device 100, by operating the control unit 1 as these functional blocks, a three-dimensional structure of a porous body having desired characteristics can be determined.

[0019] The structure generation unit 11 sets values of generation parameters for generating a porous body to be evaluated. The generation parameters are physical quantities that can be adjusted in the manufacturing process when actually manufacturing the porous body. For example, they include the average and standard deviation of the particle diameters of spheres and ellipsoids contained in the base material that is the raw material of the porous body, the volume ratio of spheres and ellipsoids in the base material, the amount and particle diameter of the pore-forming material mixed with the base material to form voids in the porous body, the sintering degree of the base material, and the like. The structure generation unit 11 can set the values of these generation parameters based on, for example, the operation content of the user input via the operation input device 4. The values of each generation parameter set by the structure generation unit 11 are stored in the storage unit 2 as generation parameter data 21.

[0020] Furthermore, based on the values of each generation parameter set as described above, the structure generation unit 11 performs a structure generation process of virtually generating a three-dimensional structure of the porous body to be evaluated on the porous body design device 100. For this process, for example, GeoDict, which is microstructure simulation software developed by Math2Market GmbH, can be used. The data representing the three-dimensional structure of the porous body generated by the structure generation unit 11 performing the structure generation process is stored in the storage unit 2 as porous body structure data 22.

[0021] The model creation unit 12 creates a learning model that can predict the characteristics corresponding to the three-dimensional structure of the porous body generated by the structure generation unit 11. A learning model is a prediction model that learns the relationship between the three-dimensional structure and characteristics of a porous body, and is also called a surrogate model. For example, using a support vector machine (SVM), support vector regression capable of predicting the characteristics of a porous body from the feature amounts of the three-dimensional structure of the porous body can be generated as a learning model. The method for constructing a learning model using SVM is well-known. For example, it is described in "Machine learning and data-driven characterization framework for porous materials: Permeability prediction and channeling defect detection" by Tomoki Yasuda, Shinichi Ookawara, Shiro Yoshikawa, and Hideyuki Matsumoto, Chemical Engineering Journal 420 (2021) 130069 DOI: 10.1016 / j.cej.2021.130069. The information of the learning model created by the model creation unit 12 is stored in the storage unit 2 as learning model data 23. Note that instead of the porous body generated by the structure generation unit 11, a learning model created by the model creation unit 12 in the past for another porous body may be used as the learning model data 23. Alternatively, a learning model acquired from the outside may be stored in the storage unit 2 as the learning model data 23. <00001 (This seems to be an error in the original. Assuming it should be

[0022] and translated as such)>0000101 Before the model creation unit 12 starts creating the learning model, the structure generation unit 11 generates a predetermined number of samples of the three-dimensional structure of the porous material. The various structural characteristics of each sample are then represented by values ​​of several predetermined types of structural descriptors. Structural descriptors representing the three-dimensional structure of the porous material include, for example, porosity, solidity, specific surface area, pore diameter, particle diameter, structural uniformity of pores or solids, pore code length, and solid code length. More specifically, structural uniformity of pores or solids refers to the variance of values ​​related to pores or solids calculated for each region after dividing the porous material into multiple regions. Examples of values ​​related to pores or solids calculated for each region include the porosity, solidity, specific surface area, pore diameter, and particle diameter mentioned above. Pore code length is the continuous length of pores in a predetermined direction within the porous material, and solid code length is the continuous length of solids in a predetermined direction within the porous material. The model creation unit 12 selects several structural descriptors from these structural descriptors that have a high contribution to the properties of the porous material as input parameters for the learning model. Then, using the values ​​of each input parameter for each sample and the properties of the porous material obtained for each sample by the following physical simulation performed by the physical simulation unit 13, it creates a learning model that predicts properties from the input parameters.

[0023] The physical simulation unit 13 performs physical calculations based on the three-dimensional structure of the porous body generated by the structure generation unit 11, and conducts a physical simulation of what happens when a predetermined substance passes through the porous body. This allows for the calculation of values ​​such as the permeability of the fluid and the filtration efficiency (collection efficiency) of particulate matter (PM) contained in the fluid when, for example, exhaust gas from an internal combustion engine passes through the porous body, as characteristics of the porous body. Furthermore, mechanical strength characteristics, electrochemical characteristics, thermal conductivity characteristics, heat exchange characteristics, electrical conductivity characteristics, gas adsorption characteristics, gas purification performance, catalyst coating properties, and the removal efficiency of substances collected in the porous body may also be included in the characteristics of the porous body to be calculated. The results of the porous body characteristics calculation by the physical simulation unit 13 are stored in the storage unit 2 as characteristic data 25. The physical simulation unit 13 can be implemented, for example, by utilizing the functions of GeoDict mentioned above.

[0024] The AI ​​calculation unit 14 uses the learning model created by the model creation unit 12 to perform calculations related to predicting the properties of the porous material using AI (Artificial Intelligence). By inputting the values ​​of each input parameter obtained for the three-dimensional structure of the porous material into the learning model, the AI ​​calculation unit 14 can perform AI calculations using well-known AI methods such as support vector regression, and predict the properties of the porous material. The values ​​of each input parameter used in the AI ​​calculation unit 14 and the obtained porous material property prediction results are stored in the storage unit 2 as input parameter data 24 and property data 25, respectively.

[0025] The characteristic evaluation unit 15 performs evaluation processing to evaluate the characteristics of the porous material calculated or predicted by the physical simulation unit 13 or the AI ​​calculation unit 14. The characteristic evaluation unit 15 performs evaluation processing of the characteristics of the porous material, for example, by plotting each calculated or predicted characteristic value on a graph. However, other methods may be used as long as the evaluation of the characteristics of the porous material can be performed appropriately. For example, an evaluation score for the characteristics of the porous material may be calculated from each characteristic value, or it may be determined whether each characteristic value meets a predetermined standard value and an OK / NG judgment on the characteristics of the porous material may be made based on the result of that determination. The characteristic evaluation results of the porous material by the characteristic evaluation unit 15 are stored in the storage unit 2 as characteristic evaluation data 26.

[0026] The gene evolution unit 16 performs a process to change the values ​​of the generation parameters. The gene evolution unit 16 performs a gene evolution process using well-known computational methods such as genetic algorithms, thereby changing the values ​​of each generation parameter represented by the generation parameter data 21 so that the properties of the porous material generated by that parameter value are improved. By repeating this gene evolution process by the gene evolution unit 16, the contents of the generation parameter data 21 are sequentially updated, and the optimal parameter values ​​can be searched for.

[0027] The model update unit 17 updates the learning model based on the results of the physical simulation performed by the physical simulation unit 13. As will be described later, the physical simulation unit 13 performs a physical simulation based on the three-dimensional structure of the porous material each time the AI ​​calculation unit 14 performs an AI calculation a predetermined number of times. Once the physical simulation is performed by the physical simulation unit 13, the model update unit 17 uses the results to update the learning model and updates the contents of the learning model data 23. This allows the learning model to reflect the results of the physical simulation and updates the learning model data 23 so that the characteristic values ​​obtained by AI calculation from each input parameter value corresponding to the three-dimensional structure of the porous material become more accurate.

[0028] The data conversion unit 18 converts the porous structure data 22, based on the three-dimensional structure of the porous body finally determined by the structure generation unit 11, into three-dimensional shape data 27. The three-dimensional shape data 27 generated by the conversion from the porous structure data 22 by the data conversion unit 18 is stored in the storage unit 2 and, as needed, read from the storage unit 2 and provided to the outside of the porous body design device 100. For example, by inputting the three-dimensional shape data 27 into a 3D printer, the 3D printer is operated according to the three-dimensional shape data 27 to manufacture a porous body that reproduces the three-dimensional structure of the porous body represented by the porous structure data 22. This makes it possible to create a prototype of the porous body designed by the porous body design device 100. Details of the manufacturing method of the porous body using a 3D printer with the three-dimensional shape data 27 will be described later.

[0029] Next, the details of the processing performed by the control unit 1 will be explained with reference to Figure 2. Figure 2 is a flowchart showing the processing flow of a porous material design apparatus according to one embodiment of the present invention. The porous material design apparatus 100 searches for a three-dimensional structure of a porous material having desired characteristics by repeatedly executing the processing shown in the flowchart of Figure 2 by the control unit 1 in response to user input. This supports the user in designing the porous material.

[0030] When the user instructs the porous body design apparatus 100 to start designing a porous body via the operation input device 4, in step S10, the control unit 1 sets the initial values ​​of variables i and j to 0. Variable i is used to count the number of times the AI ​​calculation unit 14 has performed AI calculations between physical simulation unit 13 and the next physical simulation. Variable j is used to count the number of times the gene evolution unit 16 has performed gene evolution processing, that is, the number of generations of generation parameters set so far by gene evolution processing. While the process shown in the flowchart of Figure 2 is being executed, the values ​​of variables i and j are stored in memory 3.

[0031] In step S20, the control unit 1 performs an initial learning model generation process to set the initial state of the learning model data 23. Here, the structure generation unit 11, model creation unit 12, and physical simulation unit 13 are used to generate a large number of samples of three-dimensional porous structures, and the relationship between the values ​​of input parameters selected from the structural descriptor of each sample and their characteristics is determined. Then, by creating a learning model based on the determined relationship, the initial state of the learning model data 23 can be set. Details of the initial learning model generation process will be explained later with reference to Figure 3.

[0032] In step S30, the control unit 1 sets the initial values ​​of the porous material generation parameters using the structure generation unit 11. Here, for example, the initial values ​​of the porous material generation parameters are set by randomly changing the values ​​of each generation parameter used when generating a sample of the three-dimensional structure of the porous material in the initial learning model generation process in step S20, within a predetermined range, or according to a predetermined rule. In step S30, the initial values ​​of the generation parameters are set by randomly changing the values ​​of each generation parameter by a predetermined population number (generation number of individuals) P. For example, if P=100, 100 combinations of generation parameter values ​​are set as the initial values ​​of the generation parameters. The initial values ​​of each generation parameter set here are used as the parent generation for the generation parameters of the first generation in the gene evolution process in step S150, which will be described later. Therefore, in the following description, the initial values ​​of the generation parameters are referred to as the generation parameters of the 0th generation. Once the initial values ​​of each generation parameter set in this way are stored in the storage unit 2 as generation parameter data 21, the process proceeds to step S40.

[0033] In step S40, the control unit 1 causes the structure generation unit 11 to generate a three-dimensional structure of the porous body to be subjected to characteristic prediction based on the initial value of the generation parameter set in step S30 (when j = 0) or the generation parameter of the j-th generation set in step S150 described later (when j ≥ 1). Here, as described above, for example, using simulation software such as GeoDict, a three-dimensional structure of the porous body corresponding to each generation parameter value represented by the generation parameter data 21 is virtually generated. Thereby, a three-dimensional structure of the porous body to be subjected to characteristic prediction can be generated in the same number as the preset population number P.

[0034] In step S50, the control unit 1 causes the structure generation unit 11 to calculate the values of the input parameters corresponding to each three-dimensional structure generated in step S40. Here, as described above, for example, among the structure descriptors such as porosity, solid fraction, specific surface area, pore diameter, particle diameter, structural uniformity of pores or solids, pore code length, and solid code length, each structure descriptor preselected by the structure generation unit 11 as having a high contribution to the characteristics of the porous body is used as an input parameter of the learning model, and a combination of input parameter values for each three-dimensional structure of the porous body to be subjected to characteristic prediction generated in step S40 is calculated. After storing the input parameter values calculated for each porous body to be subjected to characteristic prediction as input parameter data 24 in the storage unit 2, the process proceeds to step S60.

[0035] In step S60, the control unit 1 determines whether the value of the current variable i is equal to the predetermined value M. If i = M, the process proceeds to step S90, and if i < M, the process proceeds to step S70. Note that the value of M represents the number of settings for performing the physical simulation and updating the learning model each time the AI calculation is executed, and an arbitrary number of 1 or more, for example, M = 10 can be set.

[0036] In step S70, the control unit 1 uses the AI ​​calculation unit 14 to perform characteristic prediction using AI calculations with a learned model based on the input parameter values ​​calculated in step S50. Here, the learning model data 23 is read to acquire the learning model, and the values ​​of the input parameters obtained from the three-dimensional structure of each porous material to be predicted are input to this learning model to perform AI calculations related to the prediction of the characteristics of the porous material. This calculates characteristic prediction values ​​for each combination of input parameters and performs characteristic prediction for each porous material to be predicted. Once the characteristic prediction values ​​obtained for each porous material to be predicted are stored in the storage unit 2 as characteristic data 25, the process proceeds to step S80.

[0037] In step S80, the control unit 1 adds 1 to the value of variable i and proceeds to step S120.

[0038] In step S90, the control unit 1 performs a physical simulation based on the three-dimensional structures generated in step S40 using the physical simulation unit 13 to calculate the characteristics of each porous material to be predicted and to calculate its characteristic value. After storing the calculated characteristic values ​​obtained for each porous material to be predicted as characteristic data 25 in the storage unit 2, the process proceeds to step S100.

[0039] In step S100, the control unit 1 updates the learning model represented by the learning model data 23 based on the results of the physical simulation performed in step S90, using the model update unit 17. Here, the input parameter data 24 stored in step S50 and the characteristic data 25 stored in step S90 for the porous materials that were the subject of the physical simulations performed up to that point are read from the storage unit 2, and the contents of the learning model (weight values) are updated based on the relationship between the input parameter values ​​and characteristic values ​​for each porous material whose characteristics are to be predicted in this data. After recording the contents of the updated learning model in the learning model data 23, the process proceeds to step S110.

[0040] In step S110, the control unit 1 resets the value of variable i to 0 and proceeds to step S120.

[0041] In step S120, the control unit 1 uses the characteristic evaluation unit 15 to plot the characteristic values ​​predicted or calculated in step S70 or S90 on a graph for each porous material whose characteristics are to be predicted, and evaluates the characteristic prediction / calculation results. Here, for example, a two-dimensional graph is prepared in which the fluid permeability, one of the characteristics of the porous material, is represented on one axis, and the filtration efficiency, another characteristic of the porous material, is represented on the other axis. The characteristic values ​​predicted or calculated so far for each porous material whose characteristics are to be predicted are then plotted on this two-dimensional graph. This allows the evaluation of the characteristic prediction / calculation results so far for each porous material whose characteristics are to be predicted to be evaluated based on the position of each point plotted on the two-dimensional graph. Note that it is not necessarily required to actually plot the points on the graph, as long as the evaluation of the characteristic prediction / calculation results can be performed, and it is sufficient to simply retain data indicating the position of the points on the graph. Alternatively, as described above, the evaluation of the characteristic prediction / calculation results may be performed by a method other than plotting on a graph.

[0042] In step S125, the control unit 1 selects the next generation parent generation parameters based on the evaluation results of the characteristic prediction / calculation results obtained in step 120. Here, for example, the characteristic values ​​predicted or calculated in the previous and current steps S70 or S90 are relatively evaluated based on the positions of each point plotted on the graph in the previous and current steps S120. Then, a predetermined proportion of suitable characteristic values ​​is selected from these, for example, P characteristic values ​​(where P is the population number) corresponding to the top 50%, and the porous material generation parameters corresponding to these characteristic values ​​are selected as the next generation parent generation parameters, i.e., the (j+1) generation parent generation parameters. However, if the process in step S125 is performed for the first time, i.e., j=0, there are no previous plotted points. In this case, all of the initial values ​​of the generation parameters set in step S30 should be selected as the first generation parent generation parameters.

[0043] In step S130, the control unit 1 determines whether the value of the current variable j is equal to a predetermined value N. If j = N, the process proceeds to step S140. If j < N, the process proceeds to step S150. Note that the value of N represents the number of times the gene evolution unit 16 executes the gene evolution process, and any number of 1 or more can be set, for example, N = 100.

[0044] In step S140, the control unit 1 determines the final three-dimensional structure of the porous body based on the evaluation results of the characteristics of the porous body by the generation parameters of each generation obtained so far. Here, for example, from the positions of each point plotted on the graph in the immediately preceding step S120, the characteristic values predicted or calculated in the current step S70 or S90 are relatively evaluated, and the characteristic value with the highest evaluation is specified. The specific method will be described later with reference to FIG. 4.

[0045] If the characteristic value with the highest evaluation can be specified in the process of step S140, the control unit 1 determines the three-dimensional structure of the porous body corresponding to the characteristic value as the final three-dimensional structure of the porous body, and ends the process shown in the flowchart of FIG. 2.

[0046] In step S150, the control unit 1 sets the next candidate generation parameters using the gene evolution unit 16. Here, the next candidate generation parameters can be set by performing a gene evolution process based on the generation parameter values ​​selected as the next generation parent generation parameters in the previous step S125, and setting the child generation parameters for those parent generation parameters. Specifically, for example, using a well-known calculation method such as a genetic algorithm, the generation parameters of the j-th generation selected in step S125 are multiplied together, and a new generation parameter corresponding to its child is set as the generation parameter of the (j+1)th generation. By repeatedly performing this operation for multiple combinations of parent generation parameters, any number of next-generation child generation parameters can be generated. For example, P pairs of parent generation parameters of the j-th generation are set (P is the population number), and P child generation parameters of the (j+1)th generation are created from these pairs as the next candidate generation parameters. After recording the newly obtained generation parameter values ​​of the (j+1)th generation in the gene evolution process in the generation parameter data 21, the process proceeds to step S160.

[0047] In step S160, the control unit 1 adds 1 to the value of variable j and returns the process to step S40.

[0048] After returning from step S160 to step S40, the control unit 1 repeats the processing from step S40 onward. This generates a new three-dimensional structure of the porous material to be subject to characteristic prediction based on the generation parameters after the gene evolution processing, and predicts or calculates characteristic values ​​using AI calculations or physical simulations. Then, by adding a plot of the obtained characteristic prediction / calculation results to the graph, the control unit 1 obtains new evaluation results for the characteristic values ​​of the porous material based on each generation parameter of the next generation, in addition to the evaluation results for the characteristic values ​​of the porous material based on each generation parameter of the previous generation obtained in the previous processing. From the obtained evaluation results, the control unit 1 selects the next generation parent generation parameters to be used in the gene evolution processing. Through this series of processes, an optimization process is realized in which the optimal generation parameters are searched for by changing the generation parameters of the porous material.

[0049] Figure 3 is a flowchart detailing the initial training model generation process performed in step S20 of Figure 2.

[0050] In step S210, the control unit 1 sets the initial generation parameters for the porous body using the structure generation unit 11. Here, the initial generation parameters are set according to, for example, a predetermined combination of generation parameter values.

[0051] In step S220, the control unit 1 uses the structure generation unit 11 to generate a three-dimensional porous structure to be used for generating the initial learning model, based on the initial generation parameters set in step S210. Here, the three-dimensional structure for generating the initial learning model is generated by the same process as in step S40 in Figure 2.

[0052] In step S230, the control unit 1 uses the structure generation unit 11 to calculate the values ​​of structural descriptors corresponding to the three-dimensional structure generated in step S220. Here, for various predetermined structural descriptors such as porosity, solidity, specific surface area, pore diameter, particle diameter, structural uniformity of pores or solids, pore code length, and solid code length, the control unit 1 calculates the values ​​of each structural descriptor for the three-dimensional structure of the porous body generated in step S220.

[0053] In step S240, the control unit 1 uses the physical simulation unit 13 to perform a physical simulation based on the three-dimensional structure generated in step S220. Here, similar to the process in step S90 in Figure 2, the properties of a porous body having a three-dimensional structure for generating an initial learning model are calculated by physical simulation, and the characteristic values ​​obtained as simulation results are acquired.

[0054] In step S250, the control unit 1 determines whether each process in steps S210 to S240 has been performed for a predetermined number of samples. If the processes in steps S210 to S240 have been performed for the predetermined number of porous materials and characteristic values ​​have been calculated for each sample, the process proceeds to step S260. On the other hand, if the number of 3D structures for which characteristic values ​​have been calculated by the processes in steps S210 to S240 is less than the predetermined number of samples, and the characteristic values ​​obtained so far are less than the number of samples, the process returns to step S210 and the processes from step S210 onward are performed again.

[0055] In step S260, the control unit 1, using the structure generation unit 11, selects structural descriptors from among the structural descriptors of each sample calculated in step S230 that have a high correlation with the porous material characteristic values ​​obtained by physical simulation in step S240, as input parameters for the learning model. Here, for example, the correlation between the calculated structural descriptor value and the characteristic value for each sample is determined for each structural descriptor, and a predetermined number of structural descriptors are selected as input parameters for the learning model in order of the highest obtained correlation values. This makes it possible to select structural descriptors that have a high contribution to the characteristics of the porous material as input parameters for the learning model.

[0056] In step S270, the control unit 1 uses the model creation unit 12 to create an initial learning model based on the input parameters selected in step S260 and the characteristic values ​​of the porous material obtained in step S240. Here, the initial learning model can be created by performing model learning processing using, for example, a well-known method, based on the relationship between the input parameters and characteristic values ​​of each sample.

[0057] Once the initial learning model is created in step S270, the control unit 1 records the contents of the initial learning model in the learning model data 23 and terminates the initial learning model generation process shown in the flowchart of Figure 3. After that, the process proceeds to step S30 in Figure 2.

[0058] Figure 4 is a diagram illustrating a comparison between the three-dimensional structure of a porous material obtained by the porous material design apparatus 100 of this embodiment and the three-dimensional structure of a porous material according to a comparative example. In the graph of Figure 4, the horizontal axis represents the fluid permeability, which is one of the characteristics of a porous material, and the vertical axis represents the filtration efficiency, which is another characteristic of a porous material. In the graph of Figure 4, the values ​​of the permeability on the horizontal axis are normalized so that the minimum value is 0 and the maximum value is 1. In the graph of Figure 4, each point represented by point 43 (each point indicated by a ● marker; hereinafter referred to as "example characteristic point") shows an example of characteristic values ​​for the three-dimensional structure of a porous material according to this embodiment. These example characteristic points were obtained by repeatedly performing characteristic value calculation and optimization processing according to the flowchart in Figure 2. On the other hand, each point represented by point 41 (each point indicated by a ■ marker; hereinafter referred to as "physical calculation characteristic point") shows an example of characteristic values ​​calculated by creating the same number of three-dimensional structures of porous materials as the example characteristic points in advance and performing physical simulations for each three-dimensional structure as a comparative example. Furthermore, each point, represented by point 42 (indicated by the × marker; hereinafter referred to as "AI computation characteristic points"), shows an example of characteristic prediction values ​​obtained solely through AI computation using an initial learning model, as another comparative example. These AI computation characteristic points were obtained by always executing the process in step S70 in the flowchart of Figure 2, regardless of the value of variable i.

[0059] Comparing the points plotted on the graph in Figure 4, the AI ​​calculation characteristic point including point 42 and the embodiment characteristic point including point 43 are both located further to the upper right from the graph origin compared to the physical calculation characteristic point including point 41. In other words, the porous materials corresponding to these characteristic points have a three-dimensional structure with better characteristics than the porous materials corresponding to the physical calculation characteristic point. As will be described later, when physical simulation and optimization processing are combined, it is necessary to perform physical simulation for each of the numerous three-dimensional structures of porous materials obtained during the optimization process, which is impractical due to the enormous amount of computation time required. On the other hand, the porous material design apparatus 100 of this embodiment can predict characteristic values ​​for the numerous three-dimensional structures of porous materials obtained during the optimization process using AI calculations in a significantly shorter time than physical simulation. Therefore, it is possible to determine the three-dimensional structure of porous materials that yield good characteristics, such as the AI ​​calculation characteristic point and the embodiment characteristic point, in a realistic amount of computation time.

[0060] The AI ​​calculation characteristic points shown in Figure 4 represent the predicted characteristic values ​​obtained when the AI ​​calculation is repeatedly performed using the initial learning model generated in the initial learning model generation process in Figure 3. Therefore, there is a high possibility that the deviation from the true characteristic values ​​for the three-dimensional structure of the porous material obtained during the optimization process will be large, and it may not necessarily represent the three-dimensional structure of the porous material that truly yields good characteristics. Furthermore, the deviation from the true characteristic values ​​may prevent the correct evaluation of the superiority / inferiority relationship of the characteristic values ​​of each three-dimensional structure. On the other hand, in the porous material design apparatus 100 of this embodiment, as described above, physical simulations are performed at predetermined intervals, and the learning model is updated based on the results. Therefore, the deviation from the true characteristic values ​​for the three-dimensional structure of the porous material obtained during the optimization process is small. The example characteristic points shown in Figure 4 are the values ​​after the physical simulation update. Therefore, the example characteristic points can reliably determine the three-dimensional structure of the porous material that yields good characteristics in terms of the combination of filtration efficiency and transmittance more accurately than the AI ​​calculation characteristic points.

[0061] As described above, the porous body design apparatus 100 of this embodiment can obtain a three-dimensional porous body structure with better characteristics than the comparative example by performing the processing shown in the flowchart of Figure 2.

[0062] In the flowchart of Figure 2, the characteristic points for each embodiment shown in Figure 4 are finally obtained by repeatedly performing the process in step S120, which evaluates the characteristic values ​​predicted or calculated by AI calculation or physical simulation, for example by plotting them on a graph, and the gene evolution process in step S150. In step S140, the point showing the best characteristics among these characteristic points for each embodiment, for example, the point where both the normalized transmittance and filtration efficiency values ​​are 0.2 or higher, and which is located furthest from the graph origin, is selected as the characteristic value with the highest evaluation. Then, by identifying the three-dimensional structure corresponding to this characteristic value, the final three-dimensional structure of the porous material is determined.

[0063] For example, we compare the characteristic values ​​of each point selected from the graph in Figure 4 for each of the physical calculation characteristic points, AI calculation characteristic points, and example characteristic points, with the search time and number of individuals searched to find each of these points from the physical calculation characteristic points, AI calculation characteristic points, and example characteristic points, respectively. The comparison results are as follows, for example. In the following explanation, physical calculation characteristic points are calculated for each of the 524 three-dimensional structures of porous materials, and based on the calculation results of 500 of these, the optimization process and AI calculation or physical simulation are repeated 10 times until the total number of individuals searched reaches 5000, thereby providing an example of the characteristic values ​​and search time when AI calculation characteristic points and example characteristic points are obtained. In this example, the characteristic value and search time of the point selected from the 524 physical calculation characteristic points are set to 100 each, and the characteristic value and search time of each selected point for the AI ​​calculation characteristic points and example characteristic points are normalized and expressed accordingly. Here, a larger characteristic value indicates better characteristics, and a larger search time indicates that the processing takes longer. (1) Physical calculation characteristic points Attribute value = 100, Search time = 100 (Total time = 100), Number of individuals = 524 (2)AI calculation characteristic points Attribute value = 147, Search time per run = 6 (Total time = 56), Number of individuals searched = 5000 (3) Characteristics of the Examples Attribute value = 152, Search time per run = 15 (Total time = 150), Number of individuals searched = 5000

[0064] From the comparison results above, it can be seen that the characteristic points of the embodiment yield better characteristics than the physical calculation characteristic points and machine learning characteristic points, even with a relatively short search time. It should be noted that since the physical calculation characteristic points represent characteristic values ​​calculated by physical simulation, similar to the AI ​​calculation characteristic points and embodiment characteristic points, if 5000 physical calculation characteristic points are calculated by performing physical simulations for 5000 different three-dimensional structures of porous materials, and points with good characteristics are selected from among them, it is possible to obtain better characteristics than the embodiment characteristic points. However, this would require approximately 10 times the search time shown in the example of the physical calculation characteristic points, i.e., 1000 hours in total after normalization, which is not practical. Therefore, by adopting the porous material design method using the porous material design apparatus 100 of this embodiment, characteristic values ​​can be obtained for a large number of search individuals within a realistic timeframe, and the three-dimensional structure of a porous material with the desired characteristics can be determined in a relatively short time.

[0065] Next, the details of the manufacturing method for porous materials using a 3D printer with 3D shape data 27 will be explained with reference to Figure 5. Figure 5 is a flowchart showing the flow of the manufacturing process for porous materials using a 3D printer.

[0066] When a user instructs the porous material design apparatus 100 to start manufacturing a porous material via the operation input device 4, in step S310, the control unit 1 creates three-dimensional shape data 27 using the data conversion unit 18. Here, the porous material structure data 22 is read from the storage unit 2 and converted into three-dimensional shape data 27 by performing a predetermined data conversion process. Alternatively, the porous material structure data 22 may be used directly as the three-dimensional shape data 27. In that case, the control unit 1 does not need to have the data conversion unit 18.

[0067] In step S320, the 3D shape data 27 created in step S310 is input to a 3D printer (not shown). Here, for example, the porous material design device 100 and the 3D printer are connected wirelessly or via a wired connection, and the 3D shape data 27 is transmitted from the porous material design device 100 to the 3D printer, thereby inputting the 3D shape data 27 to the 3D printer. Alternatively, the 3D shape data 27 may be input to the 3D printer by transferring it from the porous material design device 100 to the 3D printer via a storage medium such as a USB memory stick. In addition to these, it is possible to input the 3D shape data 27 to the 3D printer by any other method.

[0068] In step S330, the 3D printer starts injecting material to form a porous body based on the 3D shape data 27 input in step S320. Then, in step S340, the 3D printer head is moved to the coordinate position represented by the 3D shape data 27 input in step S320, and then the material is injected.

[0069] In step S350, it is determined whether material has been injected from the 3D printer for all three-dimensional structures of the porous body represented by the three-dimensional shape data 27 input in step S320. If there are any three-dimensional structural parts from which material has not yet been injected, the process returns to step S340 to continue injecting material. If material has been injected for all three-dimensional structures, the process proceeds to step S360.

[0070] In step S360, the material injection in the 3D printer is terminated. This completes the manufacturing of the porous body using the 3D printer, and a porous body that reproduces the three-dimensional structure designed by the porous body design device 100 can be created. After executing the process in step S360, the process shown in the flowchart of Figure 5 is terminated.

[0071] In addition, although the above describes an example of a method for manufacturing a porous body using a 3D printer with 3D shape data 27, a porous body that reproduces the 3D structure designed by the porous body design device 100 may be manufactured by other methods. Alternatively, a porous body may be manufactured using something other than a 3D printer. If a porous body that reproduces the 3D structure designed by the porous body design device 100 can be appropriately manufactured using 3D shape data 27, any method can be selected from various known methods and used for the manufacture of the porous body.

[0072] According to the embodiments of the present invention described above, the following effects and advantages are achieved.

[0073] (1) The porous material design method using the porous material design apparatus 100 involves having the porous material design apparatus 100 perform the following multiple times: a structure generation process (step S40) in which a three-dimensional structure of the porous material is virtually generated on the porous material design apparatus 100 based on generation parameter values ​​for generating the porous material; a characteristic prediction / calculation process (steps S70, S90) in which the characteristics of the porous material having the three-dimensional structure generated by the structure generation process are predicted or calculated; an evaluation process (step S120) in which the characteristics of the porous material predicted or calculated by the characteristic prediction / calculation process are evaluated; and an optimization process (steps S125, S150) in which the optimal generation parameter values ​​are searched by changing the generation parameter values. Then, the three-dimensional structure of the porous material is determined based on the evaluation results of the characteristics of the porous material obtained by the evaluation process (step S140). In this way, a useful technology can be provided that enables the determination of the optimal structure early in the design of a porous material.

[0074] (2) In the characteristic prediction / calculation process, at least one of the following is performed: a physical simulation (step S90) that calculates the characteristics of the porous material by physical calculations based on the 3D structure generated by the structure generation process, and an AI calculation (step S70) that predicts the characteristics of the porous material using a learning model of the 3D structure. In this way, physical simulations and AI calculations can be performed as appropriate to determine a 3D structure of the porous material that yields better characteristics.

[0075] (3) In the porous body design method, the porous body design apparatus 100 is made to perform an update process (step S100) to update the learning model based on the results of the physical simulation. In this way, the learning model can be optimized to obtain more accurate characteristic prediction values ​​compared to when AI calculations using the learning model are performed alone.

[0076] (4) In the characteristic prediction / calculation process, the physical simulation in step S90 is performed every time the AI ​​calculation in step S70 is executed a predetermined number of times M (step S60: Yes). When the physical simulation in step S90 is performed in the characteristic prediction / calculation process, the porous body design apparatus 100 is made to execute the update process in step S100. In this way, the value of M can be arbitrarily changed to adjust the update frequency of the learning model, and the optimal structure can be determined in a short time and with high accuracy.

[0077] (5) The porous material can also be manufactured using the three-dimensional shape data based on the three-dimensional structure of the porous material determined by the porous material design method described above (steps S330 to S360). In this way, a prototype of the porous material designed by the porous material design apparatus 100 can be quickly and easily produced.

[0078] It should be noted that the present invention is not limited to the embodiments described above, and can be implemented using any components without departing from the spirit of the invention.

[0079] Furthermore, in the above embodiment, one of the structural descriptors representing the three-dimensional structure of the porous material is selected as an input parameter to the learning model, and the relationship between this input parameter and the properties of the porous material is represented by the learning model. However, instead of structural descriptors, generation parameters may be used as input parameters to the learning model. In this case, all of the multiple types of generation parameters that exist for the generation of the porous material may be used as input parameters, or, from among the multiple generation parameters, those that have a high contribution to the properties of the porous material may be selected as input parameters. In this way, the properties of the porous material can be directly predicted from the generation parameters by machine learning without having to perform the process of generating the three-dimensional structure of the porous material and calculating the input parameters (the process in steps S40 and S50 in Figure 2), thus making it possible to further reduce processing time.

[0080] The embodiments and modifications described above are merely examples, and the present invention is not limited to these, as long as the features of the invention are not impaired. Furthermore, although various embodiments and modifications have been described above, the present invention is not limited to these. Other embodiments conceivable within the scope of the technical idea of ​​the present invention are also included within the scope of the present invention. [Explanation of Symbols]

[0081] 1...Control unit, 2...Storage unit, 3...Memory, 4...Operation input device, 5...Display device, 6...Bus, 11...Structure generation unit, 12...Model creation unit, 13...Physical simulation unit, 14...AI calculation unit, 15...Characterization unit, 16...Gene evolution unit, 17...Model update unit, 18...Data conversion unit, 21...Generated parameter data, 22...Porous structure data, 23...Learning model data, 24...Input parameter data, 25...Characteristic data, 26...Characterization data, 27...3D shape data, 100...Porous design device

Claims

1. A computer-aided design method for porous materials, A structure generation process that virtually generates the three-dimensional structure of the porous material on the computer based on generation parameter values ​​for generating the porous material, An input parameter calculation process that selects at least one of the structural descriptors representing the three-dimensional structure as an input parameter for the learning model of the three-dimensional structure, and calculates the value of each selected input parameter. The values ​​of each input parameter calculated by the input parameter calculation process are input to the learning model, and an AI calculation process is performed to predict characteristic values ​​for a predetermined characteristic among the multiple characteristics of the porous material for the three-dimensional structure. A physical simulation process that calculates the characteristic values ​​based on the three-dimensional structure, An update process updates the learning model based on the relationship between the values ​​of each input parameter obtained by the input parameter calculation process and the characteristic values ​​obtained by the physical simulation process. An evaluation process that evaluates the characteristic value predicted by the AI ​​calculation process or calculated by the physical simulation process, The computer is made to perform an optimization process multiple times, which involves changing the generation parameter value based on the evaluation result of the characteristic value obtained by the evaluation process, in order to search for the optimal generation parameter value. Based on the evaluation results of the characteristic values ​​obtained through the evaluation process described above, the three-dimensional structure of the porous body is determined. A porous body design method in which the computer executes the physical simulation process and the update process each time the AI ​​calculation process is executed a predetermined number of times.

2. In the porous body design method according to Claim 1, The aforementioned characteristic values ​​include fluid permeability and particulate matter filtration efficiency, as particulate material design method.

3. In the porous body design method according to Claim 1, The aforementioned learning model is a porous material design method that uses a support vector machine regression model.

4. In the porous body design method according to Claim 1, The update process is a porous body design method in which the weight values ​​of the learning model are updated based on the relationship between the values ​​of each input parameter obtained by the input parameter calculation process up to the time of execution of the physical simulation process and the characteristic values ​​obtained by the physical simulation process.

5. A method for manufacturing a porous body, comprising manufacturing the porous body using three-dimensional shape data based on the three-dimensional structure of the porous body determined by the porous body design method described in any one of claims 1 to 4.

Images