Morphological search device, morphological search method and program

Abstract

To provide a morphology search apparatus, a morphology search method, and a program therefor which allows efficient search of morphology in response to a desired objective relating to a material including a plurality of construction materials.SOLUTION: A morphology data generating unit 24 generates morphology data indicating morphology of a material. A partial physical amount specifying unit 26, based on the morphology data, specifies a partial physical amount representing properties of each of a plurality of material parts each forming a part of the material. An entire physical amount specifying unit 30, based on the partial physical amount specified with respect to the plurality of material parts, specifies an entire physical amount representing properties of the entire material. A morphology search unit 32 searches morphology of the material corresponding to a target by updating morphology of at least one material part indicated by the morphology data if the entire physical amount does not satisfy a predetermined decision condition corresponding to the target.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a morphological searching device, a morphological searching method, and a program. [Background technology]

[0002] In order to clarify the relationship between the morphology of a material containing multiple components (for example, the filler morphology of filler-filled rubber) and the physical quantities that represent the characteristics of the material, molecular dynamics simulations are being conducted that model the morphology of the material, such as the aggregate structure of the fillers.

[0003] As an example of a technique related to molecular dynamics simulation, Patent Document 1 describes a time-series data generating device that generates a probabilistic time evolution model modeled by a conditional probability distribution by executing a molecular dynamics simulation.

[0004] Furthermore, Patent Document 2 describes a polymer material simulation device that can clarify the cause of breakage of a polymer material by performing a simulation using a molecular dynamics method. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] JP 2019-91237 A [Patent Document 2] JP 2019-179310 A Summary of the Invention [Problem to be solved by the invention]

[0006] However, searching for a morphology that achieves a desired goal for a material expressed in a large-scale model using molecular dynamics simulation requires an enormous amount of calculation. For example, if searching for a morphology according to the goal by repeatedly slightly shifting the position of the filler expressed in the model and recalculating the physical quantities that represent the overall properties of the material, it will take an enormous amount of time.

[0007] The present invention has been made in consideration of the above-mentioned situation, and one of its objects is to provide a morphology search device, a morphology search method, and a program that can efficiently search for morphology according to a desired goal for a material containing multiple ingredients. [Means for solving the problem]

[0008] In order to solve the above problem, the morphology search device of the present invention is a morphology search device that searches for a morphology corresponding to a desired goal for a material containing multiple materials, and includes: a morphology data generation means that generates morphology data indicating the morphology of the material; a partial physical quantity identification means that identifies, for multiple material parts that are each a part of the material, partial physical quantities that indicate characteristics of the material parts based on the morphology data; an overall physical quantity identification means that identifies an overall physical quantity that indicates an overall characteristic of the material based on the partial physical quantities identified for the multiple material parts; and a morphology search means that searches for the morphology of the material corresponding to the goal by updating the morphology of at least one of the material parts indicated by the morphology data when the overall physical quantity does not satisfy a predetermined judgment condition corresponding to the goal.

[0009] In one aspect of the present invention, the morphological data generation means generates the morphological data that satisfies given constraints relating to the material, and the morphological search means updates the morphology of at least one of the material parts indicated by the morphological data so as to satisfy the constraints.

[0010] In this aspect, the constraint condition may be a condition relating to at least one of the volume fraction of each phase, the number of phases or the number of inclusions, and the density distribution of the inclusions.

[0011] In addition, in one aspect of the present invention, when the morphology is updated, the partial physical quantity specifying means newly specifies the partial physical quantity only for the material portion whose morphology has been updated, and the entire physical quantity specifying means newly specifies the entire physical quantity for the material portion in which the newly specified partial physical quantity is reflected.

[0012] In addition, in one aspect of the present invention, the morphology search means updates the morphology data so that the morphologies of the two material portions are swapped.

[0013] In one aspect of the present invention, the method further includes a reference data storage means for storing a plurality of reference data indicating a correspondence between the partial physical quantity and the morphology in a region corresponding to the material portion or a characteristic of the material portion according to the morphology, and the partial physical quantity identification means identifies the partial physical quantity based on the reference data.

[0014] Alternatively, the partial physical quantity identifying means identifies the partial physical quantity using a trained machine learning model that outputs output data indicating the partial physical quantity in response to input of input data.

[0015] In one aspect of the present invention, the material is a filler-filled rubber, and the morphology search means searches for a position of a filler contained in the filler-filled rubber or a variable corresponding to the position.

[0016] In one aspect of the present invention, the method further includes a parameter data generation means for generating parameter data representing the overall geometric characteristics of the material based on the morphology data in which the overall physical quantity achieves the target.

[0017] In addition, the morphology exploration method of the present invention is a morphology exploration method for searching for a morphology corresponding to a desired target for a material containing multiple materials, and includes the steps of generating morphology data indicating the morphology of the material, identifying partial physical quantities representing characteristics of multiple material parts that are each part of the material based on the morphology data, identifying an overall physical quantity representing overall characteristics of the material based on the partial physical quantities identified for the multiple material parts, and searching for the morphology of the material corresponding to the target by updating the morphology of at least one of the material parts indicated by the morphology data when the overall physical quantity does not satisfy a predetermined judgment condition corresponding to the target.

[0018] In addition, the program of the present invention causes a computer that searches for a morphology corresponding to a desired goal for a material containing multiple ingredients to execute the following steps: generating morphology data indicating the morphology of the material; identifying partial physical quantities that represent characteristics of multiple material parts, each of which is a part of the material, based on the morphology data; identifying an overall physical quantity that represents characteristics of the entire material, based on the partial physical quantities identified for the multiple material parts; and searching for the morphology of the material corresponding to the goal by updating the morphology of at least one of the material parts indicated by the morphology data if the overall physical quantity does not satisfy a predetermined judgment condition corresponding to the goal. [Brief description of the drawings]

[0019] [Figure 1] FIG. 1 is a diagram illustrating an example of a configuration of a search device according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a functional block diagram showing an example of functions of a search device according to an embodiment of the present invention. [Diagram 3] FIG. 2 is a diagram illustrating an example of an entire region. [Figure 4] FIG. 13 is a diagram illustrating an example of an extended partial region. [Diagram 5] FIG. 10 is a diagram illustrating an example of identifying a partial physical quantity using a trained machine learning model. [Figure 6] FIG. 1 is a diagram illustrating an example of learning of a machine learning model. [Figure 7] 4 is a flow diagram showing an example of a flow of processing performed by a search device according to an embodiment of the present invention. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

[0021] 1 is a diagram showing an example of the configuration of a searching device 1 according to an embodiment of the present invention. In the searching device 1 according to this embodiment, a morphology search process is executed to achieve a desired goal for a material (heterogeneous material) including a plurality of materials (e.g., inclusions and a base material). An example of a material to be the target of the search process is rubber filled with a filler (filler-filled rubber).

[0022] The search device 1 according to this embodiment is a computer such as a personal computer or a server computer, and includes, for example, a processor 10, a storage unit 12, a communication unit 14, an output unit 16, and an operation unit 18.

[0023] The processor 10 is a program-controlled device such as a CPU that operates according to a program installed in the search device 1, for example.

[0024] The storage unit 12 is, for example, a storage element such as a ROM or a RAM, a hard disk drive, etc. The storage unit 12 stores programs executed by the processor 10, etc.

[0025] The communication unit 14 is a communication interface such as a network board.

[0026] The output unit 16 is, for example, a display device such as a liquid crystal display and an audio output device such as a speaker, and displays various images and outputs various sounds according to instructions from the processor 10.

[0027] The operation unit 18 is a user interface such as a keyboard or a mouse, and receives operation input from the user and outputs a signal indicating the content of the input to the processor 10.

[0028] The searching device 1 may include an optical disk drive that reads optical disks such as DVD-ROMs and Blu-ray (registered trademark) disks, a USB (Universal Serial Bus) port, and the like.

[0029] Fig. 2 is a functional block diagram showing an example of functions implemented in the search device 1 according to this embodiment. Note that the search device 1 according to this embodiment does not need to implement all of the functions shown in Fig. 2, and functions other than the functions shown in Fig. 2 may also be implemented.

[0030] As shown in FIG. 2 , the searching device 1 according to this embodiment functionally includes, for example, a goal setting unit 20, a constraint condition setting unit 22, a morphological data generating unit 24, a partial physical quantity specifying unit 26, a reference data storage unit 28, an overall physical quantity specifying unit 30, and a morphological searching unit 32.

[0031] The goal setting unit 20 and the constraint condition setting unit 22 are implemented mainly by the processor 10 and the operation unit 18. The morphological data generating unit 24, the partial physical quantity specifying unit 26, the entire physical quantity specifying unit 30, and the morphological search unit 32 are implemented mainly by the processor 10. The reference data storage unit 28 is implemented mainly by the storage unit 12.

[0032] The above functions may be implemented by executing a program including instructions corresponding to the above functions, which is installed in the search device 1, which is a computer, on the processor 10. This program may be supplied to the search device 1 via a computer-readable information storage medium, such as an optical disk, a magnetic disk, a magnetic tape, a magneto-optical disk, or a flash memory, or via the Internet, for example.

[0033] In this embodiment, the target setting unit 20 sets a given target in a physical quantity representing the characteristics of a material including a plurality of materials. Here, the target setting unit 20 may set a target in a physical quantity representing the characteristics of a filler-filled rubber. The target setting unit 20 may also generate target data indicating a target input by a user via the operation unit 18.

[0034] Here, examples of the goal indicated by the target data include, for example, maximization of a given physical quantity, minimization of a given physical quantity, a target value to be approached for a given physical quantity, a target range for a given physical quantity, a threshold value (upper limit or lower limit) for a value indicating a given physical quantity, etc. Furthermore, examples of the given physical quantity include stress, Young's modulus, hysteresis, frequency response, viscoelasticity, energy, etc.

[0035] In this embodiment, the constraint condition setting unit 22 sets a given constraint condition related to the material for which the target is set as described above. Here, for example, the constraint condition setting unit 22 may generate constraint condition data indicating the constraint condition input by the user via the operation unit 18. Here, examples of the constraint condition include the volume fraction of each phase, the number of phases and the number of inclusions, the density distribution of the inclusions, and the blending amount of the inclusions. Examples of the constraint condition when the above-mentioned material is filler-filled rubber include the volume fraction of the filler, the number of fillers, the volume fraction of the interface phase, the density distribution of the filler, and the blending amount of the filler.

[0036] In this embodiment, the morphology data generating unit 24 generates, for example, morphology data indicating the morphology of the above-mentioned material. Here, the morphology data generating unit 24 may generate morphology data that satisfies the constraint conditions indicated by the constraint condition data generated by the constraint condition setting unit 22. Examples of the morphology data include data indicating a structural model of the material in a molecular dynamics simulation or a continuum simulation. For example, the morphology data may be data indicating the position, size, and shape of a base material (e.g., a polymer) or an inclusion (e.g., a filler) contained in the material at a molecular level. The morphology data may also be data indicating an interaction between the inclusion and the base material.

[0037] In addition, in this embodiment, for example, based on the above-mentioned morphology data indicating the morphology of a material, the morphology data generating unit 24 generates partial morphology data indicating the morphology of a material portion that is a part of the material for a plurality of material portions.

[0038] 3 is a diagram showing an example of an entire region 40 corresponding to the entire material whose morphology is expressed by morphology data. As shown in FIG. 3, the morphology data according to this embodiment may be three-dimensional data.

[0039] Then, the morphological data generating unit 24 divides the entire region 40 into a plurality of partial regions 42 .

[0040] It is desirable that each of the partial regions 42 has a shape of, for example, a cube or a rectangular parallelepiped, and that all of the partial regions 42 have the same size.

[0041] In addition, since the smaller the statistical error, the higher the accuracy of the partial region 42, it is desirable that each side length is equal to or longer than 100 nanometers. In addition, if the difference between the size of the entire region 40 and the size of the partial region 42 is small, the variation of the physical quantity specified for each partial region 42 as described below will be small, so it is desirable that the size of the partial region 42 is 80% or less of the size of the entire region 40.

[0042] Moreover, when the material contains inclusions such as fillers that have extremely high rigidity, it is desirable to set 10 or more partial regions 42.

[0043] Although FIG. 3 shows multiple partial regions 42 that do not overlap with each other, the morphology data generation unit 24 may determine multiple partial regions 42 that partially overlap with each other based on the entire region 40.

[0044] Then, the morphology data generating unit 24 identifies a region of interest that corresponds to the partial region 42 .

[0045] Here, the partial region 42 itself may be specified as the region of interest corresponding to the partial region 42 .

[0046] 4, for each partial region 42, a region that partially includes the partial region 42 may be identified as an extended partial region 44 corresponding to the partial region 42. The example in FIG. 4 shows an extended partial region 44 that has the same center position as the partial region 42 and whose vertical length, horizontal length, and height length are each a predetermined multiple greater than 1. The extended partial region 44 identified in this manner may be identified as a region of interest corresponding to the partial region 42.

[0047] In addition, for each partial region 42, the entire group of partial regions 42 (e.g., 3 x 3 x 3 = 27 partial regions 42) centered on that partial region 42 may be identified as the region of interest corresponding to that partial region 42.

[0048] In this embodiment, the partial physical quantity specifying unit 26 specifies partial physical quantities (responses) representing characteristics of a plurality of material parts that are each a part of the material, for example, based on morphology data. Examples of partial physical quantities include stress, Young's modulus, hysteresis, frequency response, viscoelasticity, energy, etc., as described above.

[0049] For example, the partial physical quantity specifying unit 26 may specify, for a plurality of partial regions 42, the morphology of a region of interest corresponding to the partial region 42 indicated by the morphology data. Then, the partial physical quantity specifying unit 26 may calculate a given physical quantity in the partial region 42 based on the morphology of the specified region of interest by executing a computer simulation such as a continuum simulation or a molecular dynamics simulation. Then, the partial physical quantity specifying unit 26 may specify the calculated physical quantity as the partial physical quantity of the partial region 42.

[0050] Then, the partial physical quantity specifying unit 26 may generate reference data including reference morphology data indicating the morphology of a region of interest corresponding to the partial region 42, and reference physical quantity data indicating physical quantities calculated for the partial region 42. Then, the partial physical quantity specifying unit 26 may store the generated reference data in the reference data storage unit 28.

[0051] Then, the partial physical quantity specifying unit 26 may specify the partial physical quantity based on the reference data. For example, the partial physical quantity specifying unit 26 may check whether or not reference data including reference morphology data indicating the morphology of a region of interest corresponding to the partial region 42 that is the target of physical quantity specification is stored in the reference data storage unit 28. If the reference data is stored in the reference data storage unit 28, the partial physical quantity specifying unit 26 may specify the physical quantity indicated by the reference physical quantity data included in the reference data as the physical quantity of the partial region 42. If the reference data is not stored in the reference data storage unit 28, the physical quantity of the partial region 42 may be specified by executing a computer simulation as described above.

[0052] Furthermore, the partial physical quantity specifying unit 26 may specify reference data including reference morphology data indicating a morphology most similar to the morphology of a region of interest corresponding to the partial region 42 that is the target of physical quantity specification. Then, the partial physical quantity specifying unit 26 may specify a physical quantity indicated by the reference physical quantity data included in the specified reference data as the physical quantity of the partial region 42.

[0053] Furthermore, the reference data may include reference feature data indicating features of a material portion of a partial region 42 corresponding to a region of interest according to the morphology of the region of interest, instead of reference morphology data indicating the morphology. Examples of features indicated by the reference feature data include the volume fraction of each phase, the number of inclusions, the density distribution of inclusions, the blending amount of inclusions, and the thickness of an interface in the partial region 42.

[0054] In the present embodiment, for example, the reference data storage unit 28 stores a plurality of reference data indicating the correspondence between a partial physical quantity and a morphology in a region corresponding to a material portion or a feature of the material portion according to the morphology. As described above, the reference data storage unit 28 may store a plurality of reference data including reference physical quantity data and reference morphology data. Also, as described above, the reference data storage unit 28 may store a plurality of reference data including reference physical quantity data and reference feature data.

[0055] As described above, every time reference data for a new morphology is generated, the generated reference data may be accumulated in the reference data storage unit 28. Furthermore, prior to the search according to this embodiment, reference data for various morphologies may be generated in advance, and the reference data may be stored in the reference data storage unit 28.

[0056] Furthermore, the partial physical quantity specifying unit 26 may specify the above-mentioned feature in the partial region 42 based on the morphology of a region of interest corresponding to the partial region 42 for which the physical quantity is to be specified. Then, the partial physical quantity specifying unit 26 may specify reference data including reference feature data indicating the specified feature. Then, the partial physical quantity specifying unit 26 may specify a physical quantity indicated by the reference physical quantity data included in the specified reference data as the physical quantity of the partial region 42.

[0057] In addition, the reference data storage unit 28 may store, instead of the reference data, a trained machine learning model 50, as exemplified in FIG. 5, which outputs output data indicating the partial physical quantities described above in response to input data.

[0058] The machine learning model 50 may be, for example, a trained machine learning model that outputs output data indicating the partial physical quantities described above in response to input of input data indicating the morphology of the region of interest. The machine learning model 50 may also be, for example, a trained machine learning model that outputs output data indicating the partial physical quantities described above in response to input of input data indicating the characteristics of the material portion of the partial region 42.

[0059] The machine learning model 50 may be one that has been previously trained using training data including learning input data corresponding to the above-mentioned reference morphology data and reference feature data, and teacher data corresponding to the above-mentioned reference physical quantity data, as shown in FIG. 6, for example.

[0060] In learning the machine learning model 50, for example, learning input data included in the training data may be input to the machine learning model 50. A difference may be identified between a value of reference output data indicating an output from the machine learning model 50 corresponding to the input and a value of teacher data included in the training data. Then, supervised learning may be executed to update parameter values ​​of the machine learning model 50 by backpropagation so that a value of a loss function associated with the identified difference is minimized.

[0061] The machine learning model 50 may be a machine learning model in which machine learning is implemented, such as, for example, AdaBoost, random forest, neural network, support vector machine (SVM), nearest neighbor classifier, or the like.

[0062] Furthermore, the above-mentioned input data indicating the morphology of the region of interest or the features of the partial region 42, and the above-mentioned learning input data corresponding to the above-mentioned reference morphology data or reference feature data may be images. In this case, the machine learning model 50 may be a convolutional neural network (CNN).

[0063] Then, the partial physical quantity specifying unit 26 may specify the partial physical quantity using the trained machine learning model 50 that outputs output data indicating the partial physical quantity in response to input of input data.

[0064] For example, the partial physical quantity specification unit 26 may input input data indicating the morphology of a region of interest corresponding to the partial region 42 for which a physical quantity is to be specified, to a trained machine learning model 50 that has learned learning input data corresponding to reference morphology data. Then, the partial physical quantity specification unit 26 may specify, as the physical quantity of the partial region 42, a physical quantity indicated by output data that is an output from the trained machine learning model 50 in response to the input.

[0065] Furthermore, for example, the partial physical quantity specification unit 26 may specify the above-mentioned feature in the partial region 42 based on the morphology of a region of interest corresponding to the partial region 42 for which the physical quantity is to be specified. Then, the partial physical quantity specification unit 26 may input input data indicating the specified feature to a trained machine learning model 50 that has trained learning input data corresponding to the reference feature data. Then, the partial physical quantity specification unit 26 may specify, as the physical quantity of the partial region 42, a physical quantity indicated by output data that is an output from the trained machine learning model 50 in response to the input.

[0066] In this embodiment, for example, the total physical quantity specifying unit 30 specifies a total physical quantity representing the overall characteristics of the material based on partial physical quantities specified for a plurality of material parts. Here, for example, the total physical quantity representing the overall characteristics of the material may be specified based on partial physical quantities specified for a plurality of partial regions 42. For example, a representative value of a plurality of partial physical quantities, such as a sum or average value of a plurality of partial physical quantities, may be specified as the value of the total physical quantity. As described above, examples of the total physical quantity include stress, Young's modulus, hysteresis, frequency response, viscoelasticity, energy, and the like. Note that the physical quantity indicated by the partial physical quantity and the physical quantity indicated by the total physical quantity may be the same or different.

[0067] In this embodiment, for example, the morphology search unit 32 searches for the morphology of the material according to the target based on the overall physical amount specified for each of at least one piece of morphology data related to the material.

[0068] The morphological search unit 32 may determine whether the total physical quantity achieves a desired target. For example, the morphological search unit 32 may determine whether the specified total physical quantity satisfies a determination condition according to the target indicated by the target data generated by the target setting unit 20.

[0069] Then, the morphology search unit 32 may search for the morphology of the material according to the desired target by updating the morphology of at least one material part indicated by the morphology data when the total physical quantity does not satisfy a predetermined judgment condition according to the desired target. Also, the morphology search unit 32 may update the morphology data so as to satisfy the constraint condition indicated by the constraint condition data generated by the constraint condition setting unit 22.

[0070] The morphological search unit 32 may also determine what kind of update to perform based on the difference between the target physical quantity and the specified overall physical quantity. As a morphological update method, for example, an optimization method such as the steepest descent method or a sampling method such as MCMC may be used. Also, a downhill simplex method, a genetic algorithm, a Newton method, a conjugate gradient method, an extended ensemble method, or the like may be used. The morphological search unit 32 may also determine what kind of update to perform by combining a plurality of the above-mentioned methods. For example, the morphological search unit 32 may perform sampling by MCMC and perform the steepest descent method using the morphology obtained by the sampling as an initial value.

[0071] Furthermore, the morphology searching unit 32 may update the morphology data so that the morphologies of the two random material portions (the morphologies shown in the two partial regions 42) are interchanged.

[0072] Furthermore, the morphology search unit 32 may specify a gradient (sensitivity) of a change in physical quantity with respect to a change in morphology for each partial region 42. Then, the morphology of the partial region 42 determined based on the specified gradient (sensitivity) may be updated. For example, the morphology of the partial region 42 whose value indicating sensitivity is greater than a predetermined value or whose value indicating sensitivity is the largest may be updated.

[0073] Here, the morphology of the partial region 42 with low sensitivity may be updated at the same time so as to satisfy the above-mentioned constraints. For example, the morphology of the partial region 42 with a sensitivity value greater than a predetermined value may be replaced with the morphology of the partial region 42 with a sensitivity value smaller than a predetermined value. Also, for example, the morphology of the partial region 42 with the highest sensitivity value may be replaced with the morphology of the partial region 42 with the lowest sensitivity value.

[0074] The reference data may also include data indicating a gradient (sensitivity) of a change in a physical quantity relative to a change in morphology. Then, the morphology searching unit 32 may update the morphology of the partial region 42 determined based on the gradient (sensitivity) indicated by the reference data.

[0075] Furthermore, the morphology search unit 32 according to the present embodiment may directly search in advance for a morphology that achieves a target related to a desired physical quantity while satisfying a given constraint condition in a region corresponding to the region of interest (a region smaller than the entire region 40). Then, the morphology search unit 32 may generate reference data including reference physical quantity data indicating the physical quantity and reference morphology data indicating the morphology that is the search result. Then, the morphology search unit 32 may store the reference data generated in this manner in the reference data storage unit 28. Then, the partial physical quantity identification unit 26 may identify the partial physical quantity as described above based on the reference data stored in the reference data storage unit 28 in this manner. Note that, in the present embodiment, it is preferable that various patterns of reference data generated by performing the above-mentioned direct search for various targets are stored in the reference data storage unit 28.

[0076] In the above-mentioned direct search of morphology, the positions of inclusions such as fillers may be expressed by continuous values. The positions of inclusions such as fillers expressed by continuous values ​​may be updated in the search. In this way, although the search takes time, it becomes possible to search for any morphology.

[0077] Furthermore, in the above-mentioned direct search of morphology, the position of an inclusion such as a filler may be expressed by any one (discrete value) of a plurality of predetermined coordinate values ​​(e.g., coordinate values ​​at regular intervals). For example, the position of an inclusion such as a filler may be searched from among coordinate values ​​expressed by integers such as 0, 1, 2, 3, . . . In addition, the position of an inclusion such as a filler may be searched from among coordinate values ​​expressed by real numbers at intervals of 0.1, such as 0, 0.1, 0.2, . . . In this way, although only limited morphologies that can be expressed by combinations of coordinate values ​​can be searched, it is possible to search a large amount of morphology space in a short time.

[0078] Here, an example of the flow of processing performed by the searching device 1 according to this embodiment will be described with reference to the flow diagram shown in FIG.

[0079] First, the target setting unit 20 generates target data indicating targets set for physical quantities that represent the properties of materials including a plurality of materials (S101).

[0080] Then, the constraint condition setting unit 22 generates constraint condition data indicating given constraint conditions related to the material for which the target is set in the process shown in S101 (S102).

[0081] Then, the morphology data generating unit 24 generates initial state morphology data that satisfies the constraint conditions indicated by the constraint condition data generated in the process shown in S102 (S103). Here, for example, morphology data that indicates an arbitrary morphology that satisfies the constraint conditions is generated.

[0082] Then, the morphology data generating unit 24 divides the entire region 40, which corresponds to the entire material whose morphology is expressed by the morphology data generated in the process shown in S103, into a plurality of partial regions 42 (S104).

[0083] Then, for each of the partial regions 42 resulting from the division in the process shown in S104, the morphology data generating unit 24 determines a region of interest corresponding to that partial region 42 (S105).

[0084] Then, for each of the multiple partial regions 42 resulting from the division in the process shown in S104, the partial physical quantity identification unit 26 identifies a partial physical quantity of the partial region 42 based on the morphology of the region of interest corresponding to the partial region 42 (S106).

[0085] Then, the entire physical quantity specifying section 30 specifies an entire physical quantity based on the partial physical quantities specified for each of the partial regions 42 in the process shown in S106 (S107).

[0086] Then, the morphological search unit 32 determines whether or not the overall physical quantity identified in the process shown in S107 satisfies a determination condition according to the target indicated by the target data generated in the process shown in S101 (S108).

[0087] If it is determined that the determination condition is not satisfied (S108: N), the morphological search unit 32 identifies the above-mentioned gradient (sensitivity) for each partial region 42 (S109).

[0088] Then, the morphology search unit 32 determines at least one partial region 42 for updating the morphology based on the gradient (sensitivity) identified for each partial region 42 in the process shown in S109 (S110).

[0089] Then, the morphology search unit 32 updates the morphology of at least one partial region 42 determined in the process shown in S110, which is indicated by the morphology data (S111), and the process returns to the process shown in S106.

[0090] If it is determined in the process shown in S108 that the determination condition is satisfied (S108: Y), the morphology search unit 32 specifies the morphology indicated by the latest morphology data as the morphology of the material corresponding to the desired target (S112). Then, the process shown in this processing example is terminated.

[0091] In this embodiment, when the morphology is updated, the partial physical quantity specifying unit 26 may newly specify partial physical quantities only for the material portion whose morphology has been updated. Then, for the material portion whose partial physical quantity has been newly specified, the entire physical quantity specifying unit 30 may newly specify an entire physical quantity reflecting the newly specified partial physical quantity.

[0092] For example, in the process shown in S106 executed for the second or subsequent times after the process shown in S111, partial physical quantities may be specified only for material portions whose morphology has been updated. Then, in the process shown in S107 executed thereafter, for material portions whose partial physical quantities have been newly specified, a total physical quantity reflecting the newly specified partial physical quantities may be newly specified.

[0093] In this way, the amount of calculation in the process performed by the searching device 1 can be reduced because the partial physical quantities are not recalculated for the material portions whose morphology has not been updated.

[0094] In the process shown in S112 of the above processing example, the morphology search unit 32 may generate parameter data representing the overall geometric characteristics of the material (i.e., the geometric characteristics of the entire region 40) based on the morphology shown by the latest morphology data. Examples of the parameter data include, in addition to indicators such as the volume fraction of each phase, functions expressing geometric shapes such as a radial distribution function, a K function, an L function, a Probability function, and a Cluster function, and data showing the results of fitting these with a polynomial or an exponential function. In this way, for example, the parameter data can be used to easily make a comparison with physical quantities based on an actual sample.

[0095] The morphology search unit 32 may search for the position of the filler contained in the filler-filled rubber or a variable corresponding to the position. Here, the variable corresponding to the position of the filler may be, for example, a variable indicating a geometric feature represented by the above-mentioned parameter data. The variable corresponding to the position of the filler may be a variable that is a parameter in a Poisson process or the like.

[0096] In this embodiment, the morphology search unit 32 may generate discrete continuous data calculated from different times or different deformation states, or data showing the difference between them. Specifically, for example, data showing a stress-strain curve or hysteresis may be generated.

[0097] In this embodiment, as described above, a partial physical quantity is specified for each material portion, and a total physical quantity is specified based on the partial physical quantities specified for the multiple material portions. Then, when the total physical quantity does not satisfy a predetermined judgment condition according to a desired target, the morphology of at least one material portion indicated by the morphology data is updated. In this way, according to this embodiment, it is possible to efficiently search for a morphology according to a desired target for a material including multiple materials.

[0098] Furthermore, by using the machine learning model 50 as described above, the accuracy of identifying morphology is improved and the calculation time required to identify the morphology is reduced compared to the case of performing computer simulation.

[0099] Furthermore, by updating the morphology taking into account the constraints as described above, for example, in the case of filler-filled rubber, the filler blend amount and aggregation structure in the actual sample are reproduced with a morphology that meets the desired target.

[0100] In addition, since the search according to this embodiment is a combinatorial search, by specifying partial physical quantities based on reference data as described above in this embodiment, it is possible to efficiently specify morphology according to a desired target. For example, by storing reference data for morphologies that are difficult to sample, such as ultra-high elasticity, in the reference data storage unit 28 in advance, it is possible to widen the range of reproducible physical quantities.

[0101] It should be noted that the present invention is not limited to the above-described embodiment.

[0102] For example, the structural model indicated by the morphology data may be a three-dimensional model as described above, or may be a two-dimensional model.

[0103] Furthermore, the specific numerical values ​​and character strings described above, as well as the specific numerical values ​​and character strings in the drawings, are merely examples, and the present invention is not limited to these numerical values ​​and character strings. [Explanation of symbols]

[0104] 1 Search device, 10 Processor, 12 Memory unit, 14 Communication unit, 16 Output unit, 18 Operation unit, 20 Goal setting unit, 22 Constraint condition setting unit, 24 Morphology data generation unit, 26 Partial physical quantity identification unit, 28 Reference data memory unit, 30 Overall physical quantity identification unit, 32 Morphology search unit, 40 Overall region, 42 Partial region, 44 Extended partial region, 50 Machine learning model.

Claims

1. A morphology exploration device for exploring a morphology according to a desired target for a polymeric material including a plurality of materials, comprising: a morphology data generating means for generating morphology data representing the morphology of the polymer material; a partial physical quantity specifying means for specifying partial physical quantities representing characteristics of a plurality of material portions that are each a part of the polymer material based on the morphology data; an overall physical quantity specifying means for specifying an overall physical quantity that represents an overall characteristic of the polymer material based on the partial physical quantities specified for the plurality of material portions; a morphology search means for searching for a morphology of the polymer material according to the target by updating a morphology of at least one of the material portions indicated by the morphology data when the total physical quantity does not satisfy a predetermined judgment condition according to the target; 13. A morphological searching apparatus comprising:

2. the morphology data generating means generates the morphology data that satisfies a given constraint condition related to the polymer material; the morphological search means updates the morphology of at least one of the material portions as indicated by the morphological data so as to satisfy the constraints; 2. The morphological searching device of claim 1.

3. The constraint condition is a condition related to at least one of the volume fraction of each phase, the number of phases or the number of inclusions, and the density distribution of inclusions.

3. The morphological searching device of claim 2.

4. When the morphology is updated, the partial physical quantity specifying means newly specifies the partial physical quantity only for the material portion whose morphology has been updated, and the entire physical quantity specifying means newly specifies the entire physical quantity in which the newly specified partial physical quantity is reflected for the material portion. A morphological searching device according to any one of claims 1 to 3.

5. The morphology search means updates the morphology data so that the morphologies of the two material portions are interchanged. Morphological searching device according to any one of claims 1 to 4.

6. a reference data storage means for storing a plurality of reference data indicating a correspondence between the partial physical quantity and the morphology in a region corresponding to the material portion or a characteristic of the material portion according to the morphology, the partial physical quantity identifying means identifies the partial physical quantity based on the reference data. Morphological searching device according to any one of claims 1 to 5.

7. the partial physical quantity identifying means identifies the partial physical quantity using a trained machine learning model that outputs output data indicating the partial physical quantity in response to input of input data; Morphological searching device according to any one of claims 1 to 5.

8. the polymeric material is a filled rubber; The morphology search means searches for a position of a filler contained in the filler-filled rubber or a variable corresponding to the position. Morphological searching device according to any one of claims 1 to 7.

9. and a parameter data generating means for generating parameter data representing an overall geometric characteristic of the polymer material based on the morphology data in which the overall physical quantity achieves the target. Morphological searching device according to any one of claims 1 to 8.

10. A morphology exploration method for exploring a morphology according to a desired target for a polymeric material including a plurality of materials, comprising: generating morphological data indicative of the morphology of the polymeric material; Identifying partial physical quantities that represent characteristics of a plurality of material portions that are each a part of the polymer material based on the morphology data; determining an overall physical quantity representing an overall characteristic of the polymeric material based on the partial physical quantities determined for the plurality of material portions; a step of searching for a morphology of the polymer material according to the target by updating a morphology of at least one of the material portions indicated by the morphology data when the total physical quantity does not satisfy a predetermined judgment condition according to the target; 16. A computer implemented morphological searching method comprising:

11. A computer that searches for a morphology according to a desired goal for a polymer material including a plurality of materials, generating morphological data indicative of the morphology of the polymeric material; a step of identifying partial physical quantities representing characteristics of a plurality of material portions, each of which is a part of the polymer material, based on the morphology data; determining an overall physical quantity representing an overall characteristic of the polymeric material based on the partial physical quantities determined for the plurality of material portions; a step of searching for a morphology of the polymer material according to the target by updating a morphology of at least one of the material portions indicated by the morphology data when the entire physical quantity does not satisfy a predetermined judgment condition according to the target; A program characterized by executing the above.

Images