A smart materials design platform inspired by living organisms.

Abstract

This invention discloses a bio-inspired smart materials design platform that satisfies the design of multi-purpose materials featuring futuristic composite microstructures. The platform sets the mechanical properties of simulated material elements by constructing a scaled-down model. The distribution of the simulated material elements is simulated to output material simulation parameters. A deep learning framework is incorporated into the platform to calculate and evaluate the optimal materials design that fits the target materials parameters. Specifically, the scaled-down model can be based on data provided by any test of the material's mechanical properties, and the deep learning framework evaluates whether the biomimetic materials design meets the requirements of the optimal target materials parameters based on a standardized reward function model. The platform is applicable to multi-purpose simulated materials design and has high potential for future applications.

Description

Technical Field

[0001] The present invention relates to material design inspired by organisms and integrates artificial intelligence technology. Specifically, deep learning is applied to the simulation of the material structure inspired by the organisms.

Background Art

[0002] With the development of technology, differences are emerging in the application of materials and structures in various fields and industries. Using a single structure or material for design and application cannot meet the requirements of complexity and precision that are fundamental in high-value-added manufacturing. The emergence of 3D printers has made it possible to manufacture composite-structure materials quickly and precisely, and such technology has been frequently applied to material research in various industries.

[0003] Taking aerospace technology as an example, the requirements for parts and components of spacecraft are continuously increasing. As the flight speed of spacecraft improves, the structural stability of parts and components is required to ensure the safety of spacecraft. In the issue of energy conservation, fuel consumption and carbon emissions depend on the mass of the spacecraft, so in material design, weight reduction, robustness, and durability are emphasized.

[0004] In machining, the requirements for high precision and high acceleration of machine tools are becoming increasingly difficult for conventional metal machine tools to meet. For the miniaturization of movable parts of machine tools, high strength and rigidity, as well as a high-precision and low-tolerance margin, are required. The material design of movable parts to meet the above requirements aims at the complexity of the strain stress faced by the movable parts.

[0005] As human civilization progresses, the sports industry is attracting increasing attention. To meet the diverse demands for sports equipment, providing sports equipment that is durable, lightweight, and comfortable has become mainstream in the sports industry. In terms of user comfort during exercise, the ability of new types of materials to absorb the impact or reaction forces generated during the course of exercise can be a critical factor. Wearable devices such as sneakers, hiking boots, swimming aids, helmets, and armor needed to withstand the stress received from the limbs during exercise, as well as the impact forces from collisions with other equipment or the ground. Because there is more than one source of impact force, previous designs using single materials and single structures had difficulty meeting the above requirements.

[0006] Currently, materials with higher "specific stiffness" or "specific strength" are being used to improve new material designs, and material distribution is being optimized by analyzing structural load density based on specific materials. In addition to structural strengthening and load density analysis in conventional materials science, material design that mimics biological structures in nature is also a rapidly growing field. Living organisms have developed biological structures with excellent mechanical properties over a long evolutionary history while facing complex and rapidly changing natural environments. The structural stability and strength of beehives, the stiffness and strength of spiderwebs, or the outstanding hydrodynamic properties and waterproofing produced by the tiny teeth on the surface of shark skin are good examples of biological structures. Biomimetic microstructures combined with 3D printers are expected to meet future industrial demands for material design with excellent functional properties including high strength, high energy absorption, or lightweight properties.

[0007] Therefore, there was an urgent need for a new, data-driven, and multidimensional finite element-simulated platform for the design of materials and composite structures. Such a platform would enable the design of composite microstructures required in a variety of properties and fields, such as applications in the aerospace, military, automotive, bulletproof coating, or sports industries. [Overview of the project] [Problems that the invention aims to solve]

[0008] To address the aforementioned technical challenges faced by biomimetic material structural design, it is necessary to overcome a wide range of parameters related to composite structures, such as porosity, porous structure, strain stress, reaction force, or strain energy. To perform structural design using a material design model capable of handling diverse parameters and the corresponding scales, this invention leverages the advantages of meta-structural models to generate multiple types of microstructures for analysis, obtain fundamental material parameters by 3D printing materials with different porous microstructures, and conducting experiments. These parameters, combined with finite element simulations, contribute to training artificial intelligence for material design.

[0009] The bio-inspired smart materials design platform disclosed in this invention comprises: a materials distribution simulation module configured to perform a target materials distribution simulation to acquire materials simulation parameters; and a reinforcement learning module configured with a deep learning framework that performs a reward function model to communicate with the materials distribution simulation module and calculate the reward values ​​of iterative materials simulants of other target materials based on an optimal target parameter (P), and evaluate whether the iterative materials simulants fit the optimal target parameter (P). Here, the target materials include rigid and soft materials, each material having a different combination of materials model coefficients, and the optimal target parameter (P) is set by reference to the materials simulation parameters, including strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more thereof.

[0010] In some embodiments, the reward function model includes modifying an initial distribution simulator to the iterative material simulator to obtain a variable quantity (D) based on the deep learning framework; assigning an initial reward value (q1) to the initial distribution simulator and an ending reward value (q2) to the iterative material simulator to calculate a standardized reward value (Q); and evaluating whether the iterative material simulator fits the optimal target parameter (P) based on the standardized reward value (Q) using the deep learning framework.

[0011] JPEG2026527427000002.jpg48151

[0012] In some embodiments, the bio-inspired smart material design platform further comprises a compression experiment module connected to the reduction model configuration module, the compression experiment module including an experimental structure generator for generating a metastructure based on a metastructure model, and a structure compressor for compressing the metastructure to obtain the compression data, wherein the metastructure is the same as or different from the experimental structure.

[0013] Preferably, the material distribution simulation module includes a finite element simulator configured to generate the initial distribution simulant by arranging the target material based on the optimal target parameter (P).

[0014] In one or more embodiments, the bio-inspired smart material design platform includes a reduced model configuration module that communicates with the material distribution simulation module to construct the target material. The reduced model configuration module includes a master curve generator that generates a master curve based on compressed data, the compressed data including Young's modulus, plateau stress, and relative density; a material coefficient curve fitting simulator that performs a curve fitting of a selected experimental structure to the master curve to obtain a combination of material coefficients and outputs a combination of material model coefficients; and a target material generator that assigns the combination of material coefficients to a single element to obtain the target material, the combination of material model coefficients including a combination of hard material model coefficients, a combination of soft material model coefficients, or a combination thereof.

[0015] Preferably, the experimental structure includes a gyroid, primitive, F-RD, Fischer-Koch S, diamond, or I-WP, or the experimental structure is generated based on a metastructure model, the metastructure model includes a triple periodic minimal surface model (TPMS model).

[0016] In a preferred embodiment, the bio-inspired smart material design platform further comprises a compression experiment module connected to the reduction model configuration module, the compression experiment module including an experimental structure generator for generating a metastructure based on a metastructure model, and a structure compressor for compressing the metastructure to obtain the compression data, wherein the metastructure is the same as or different from the experimental structure.

[0017] Preferably, the experimental structure includes a gyroid, primitive, F-RD, Fischer-Koch S, diamond, or I-WP, or the experimental structure is generated based on a metastructure model.

[0018] Preferably, the metastructure model includes a triple periodic minimal surface model (TPMS model).

[0019] In another preferred embodiment, the material distribution simulation module includes a finite element simulator configured to generate a material distribution simulant using the target material, generate a rigid compressible body, and output the material simulation parameters by simulating the compression of the rigid compressible body on the material distribution simulant.

[0020] Preferably, the deep learning framework includes Deep Q-Networks.

[0021] In various embodiments, the bio-inspired smart materials design platform further comprises a materials design module that communicates with the reinforcement learning module to design a bio-inspired simulated distribution simulant based on a second optimal target parameter customized by the user, wherein the second optimal target parameter is set by reference to the materials simulation parameters, including strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more thereof.

[0022] In other aspects, the method for designing a smart material inspired by the organisms disclosed in the present invention includes performing a target material distribution simulation by a material distribution simulation module to obtain material simulation parameters, wherein the target material includes a hard material and a soft material, and different combinations of material model coefficients are set for each material, and a deep learning framework configured as a reinforcement learning module for calculating a reward value and evaluating whether an iterative material simulator conforms to the optimal target parameter (P) executes a reward function model, and the reward value is calculated to correspond to an iterative material simulator of another target material based on the optimal target parameter (P), and the optimal target parameter is set by referring to the material simulation parameters including strain energy (SE), reaction force (RF), average von Mises stress (ST), or a combination of two or more of them.

[0023] In various embodiments, the calculation of the reward value includes modifying an initial distribution simulator to the iterative material simulator by the reward function model to obtain a variable quantity (D) based on the deep learning framework, assigning an initial reward value (q1) to the initial distribution simulator and an end reward value (q2) to the iterative material simulator to calculate a normalized reward value (Q), and evaluating whether the iterative material simulator conforms to the optimal target parameter (P) based on the normalized reward value (Q) by the deep learning framework.

[0024] Preferably, the normalized reward value (Q) is calculated based on the following formula. Q = [q1 q2] * d * P In the formula, d is a normalization coefficient calculated based on the following formula. d = σ[(D - α) / β], In the formula, α is the average value of the variable quantity (D), β is the standard deviation of the variable quantity (D), and σ is a sigmoid function.​​​In a preferred embodiment, the material distribution simulation includes generating a material distribution simulator of the target material by a finite element simulator, generating a rigid compressor, and simulating the compression of the rigid compressor that compresses the material distribution simulator, thereby exporting the material simulation parameters.

[0026] More preferably, the initial distribution simulator is generated by a finite element simulator by arranging the target material based on the optimal target parameters (P).

[0027] In another preferred embodiment, the method includes generating a master curve by a master curve generator based on compression data, the compression data including Young's modulus, plateau stress, and relative density, obtaining a combination of material coefficients, and exporting a combination of material model coefficients, performing curve fitting of a selected experimental structure to the master curve by a material coefficient curve fitting simulator, assigning a combination of the material coefficients to a single element by a target material generator to obtain the target material, and the combination of the material model coefficients including a combination of hard material model coefficients, a combination of soft material model coefficients, or a combination thereof, and further including modeling a reduced model.

[0028] Preferably, the experimental structure includes a gyroide, primitive, F-RD, Fischer-Koch S, diamond, or I-WP, or the experimental structure is generated based on a meta-structure model.

[0029] More preferably, obtaining the compression data includes generating the compression data by a compression experiment module. Generating the compression data includes generating a meta-structure based on a meta-structure model by an experimental structure generator, and compressing the meta-structure by a structure compressor to obtain the compression data, and the meta-structure is the same as or different from the experimental structure.

[0030] Preferably, the experimental structure includes a gyroid, primitive, F-RD, Fischer-Koch S, diamond, or I-WP, or the experimental structure is generated based on a metastructure model.

[0031] More preferably, the metastructure model includes a triple periodic minimal surface model (TPMS model).

[0032] In a preferred embodiment, the method further comprises designing a bio-inspired simulated distribution simulant by a material design module based on a second optimal target parameter customized by the user, wherein the second optimal target parameter is set by reference to the material simulation parameters, which include strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more thereof.

[0033] This invention provides a standardized workflow for material design. Because the reward function model of the reinforcement learning model is appropriately defined, the corresponding structural design can be effectively simulated, meeting the requirements for material structural design in various situations.

[0034] The reduced model disclosed in this invention improves the subsequent calculation of the target material distribution by setting material simulation parameters for the target material. The reduced model reduces the number of composite structure parameters involving the target material and shortens the time required to compute the reinforcement learning model.

[0035] The bio-inspired smart material design platform disclosed in this invention uses an artificial intelligence-based deep learning framework for biomimetic material design. The biomimetic material, designed with reference to simulation results, successfully reduces the concentration of reaction forces and strain stresses during the compression process. Such a platform is beneficial for material design involving complex microstructures and can be widely applied to the design of sporting goods, including shoe midsoles, integrally molded sandals, badminton racket grips, helmets, and golf balls and golf clubs. [Brief explanation of the drawing]

[0036] [Figure 1A] This is a block chart showing the structure of a bio-inspired materials design platform according to one embodiment of the present invention. [Figure 1B] This is a simulation of a shoe midsole, providing a schematic diagram of the material distribution simulator. [Figure 2A] This flowchart shows a method for designing bio-inspired smart materials according to several embodiments of the present invention. [Figure 2B] This flowchart shows a method for designing bio-inspired smart materials according to several embodiments of the present invention. [Figure 3] This document describes a specific embodiment of a method for designing bio-inspired smart materials according to one embodiment of the present invention. [Figure 4] The structure of the test specimen in the compression experiment and the resulting stress-strain curve are shown. [Figure 5A] The master curve is shown. [Figure 5B] The master curve is shown. [Figure 6] This is a flowchart showing the process of fitting a material model coefficient curve according to a specific embodiment. [Figure 7] This flowchart shows the process of fitting material simulation parameter curves according to a specific embodiment. [Figure 8] This flowchart shows the execution process of a reward function model using a deep learning framework and a materials design inspired by biological processes. [Figure 9A] This is a schematic diagram of the hard-soft material distribution in a material distribution simulator, and a simulation diagram of its strain-stress distribution. [Figure 9B] This is a schematic diagram of the hard-soft material distribution in a material distribution simulator, and a simulation diagram of its strain-stress distribution. [Figure 9C] This is a schematic diagram of the hard-soft material distribution in a material distribution simulator, and a simulation diagram of its strain-stress distribution. [Figure 10] This is a schematic diagram illustrating the technology integrated into the present invention and its corresponding future applications. [Modes for carrying out the invention]

[0037] The following examples illustrate the technical significance of the present invention and the specific technical effects achieved. This is not intended to limit the scope of protection to which it is necessary to refer to the content included in the claims. All improvements or extensions derived from the technical spirit of the present invention are also included in the scope of protection under the claims of this application.

[0038] In one embodiment, as shown in Figure 1A, the bio-inspired smart material design platform (100) according to the present invention comprises: a material distribution simulation module (1) configured to perform a target material distribution simulation to acquire material simulation parameters; and a reinforcement learning module (2) configured with a deep learning framework (21) to communicate with the material distribution simulation module and to perform a reward function model (22) to calculate the reward value of an iterative material simulant of other target materials based on an optimal target parameter (P), and to evaluate whether the iterative material simulant fits the optimal target parameter (P). The target materials include hard materials and soft materials, each material having a different combination of material model coefficients. The optimal target parameter (P) is set based on the material simulation parameters, including strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more of these.

[0039] In one or more embodiments, as shown in Figure 1A, the material distribution simulation module (1) includes a finite element simulator (11) configured to generate a material distribution simulant using the target material, generate a stiff compressible body, and simulate the stiff compressible body to compress the material distribution simulant, thereby exporting the material simulation parameters.

[0040] The aforementioned biologically inspired smart material design platform (100) further comprises a reduced model configuration module (3) that communicates with the material distribution simulation module (1) to constitute the target material. The reduced model configuration module (3) comprises a master curve generator (31) for generating a master curve based on compressed data, wherein the compressed data includes Young's modulus, plateau stress, and relative density; a material coefficient curve fitting simulator (32) that performs curve fitting of a selected experimental structure to the master curve, obtains combinations of material coefficients, and outputs combinations of material model coefficients; and a target material generator (33) that assigns the combinations of material coefficients to a single element to obtain the target material, wherein the combinations of material model coefficients include combinations of hard material model coefficients, combinations of soft material model coefficients, or combinations thereof.

[0041] Incidentally, the compression data may also be material engineering parameters of the target material corresponding to any of the experimental structures. Examples of the compression data include Young's modulus, shear modulus, elastic modulus, shear modulus, strain stress, deformation coefficient, strain energy, and strain variation coefficient. Specifically, stress refers to the biasing force applied to a unit area, and the biasing force may be tensile force, compressive force, shear force, bending force, or torsional force, but the present invention is not limited to these. Strain refers to the deformation of an article due to the biasing force, such as plastic deformation and elastic limit. The material engineering parameters mentioned above are not limited to a single parameter or a group of different parameters. The material engineering parameters may also be a ratio of two or more of the above parameters. For example, a proportional limit is defined by the correlation between stress and strain. Preferably, the compression data is obtained by the compression experiment module (4) substantially performing the compression of the experimental structure using a compression test apparatus. The experimental structure is obtained by referencing a simulation solid structure and printing it with a 3D printer. It is understood that the acquisition of the compression data is not limited to the methods described above. The materials engineering parameters corresponding to the metastructure include one of the following: a microstructure, a biomimetic microstructure, and a lattice structure, which can be used to construct the scaled-down model.

[0042] In a preferred embodiment, as shown in Figure 1A, the compression experiment module (4) is connected to the reduction model configuration module (31). The compression experiment module (4) comprises an experimental structure generator (41) for generating a metastructure based on a metastructure model, and a structure compressor (42) for compressing the metastructure to obtain the compression data, wherein the metastructure is the same as or different from the experimental structure. The metastructure refers to a cellular structure, and more specifically, a lattice structure exhibiting mechanical properties including high specific stiffness, specific strength, or impact energy absorption. Microstructural design of the lattice structure allows for tuning of the mechanical properties to meet various material design requirements. For example, tuning the coefficient of thermal expansion (CTE) and Poisson's ratio (PR) to positive, neutral, or negative can be a structural modification strategy. Specifically, the metastructure is not limited to these. Having a regular microstructure, any solid structure obtained based on a metastructure model by its mathematical algorithm is available for use by the bio-inspired platform (100) for material design according to the present invention, and basic metastructure units can be exemplified by idealized foams, kelvins, cubes, or octets.

[0043] In particular, the metastructure model includes a triple periodic minimal surface model (TPMS model). The TPMS model can be used to generate regular complex cell structures.

[0044] Incidentally, the simulation compression according to the present invention is based on a finite element model, and the finite element model may be ABAQUS, ANSYS, OpenFOAM, SimScale, Autodesk CFD, or RoboLogix, but the present invention is not limited to these. The simulation compression includes setting the center of compression of the material distribution simulant, and the center of compression may be a geometric center (known as a centroid) or any surface point of the material distribution simulant. For example, as shown in Figure 1B, an example of the material distribution simulant is a shoe midsole simulant, and its center of compression can be set based on specific requirements of the shoe midsole. For example, if the requirement is high support capacity of the arch (A), the center of compression is set at part A of the arch. In another example, if the requirement is reduced plantar vibration and a high energy absorption design, the center of compression is set at part H of the heel. The setting of the center of compression is not limited to the above examples, and the center of compression can be customized according to the user's needs. For example, the center of the simulated compression may be set by the user referring to the center of gravity or center of mass when walking, running, or standing. Alternatively, the center of the simulated compression may be set by referring to the state in which force is exerted when the heel force exerted in the aforementioned actions is stronger than the arch force. The center of the simulated compression can be set according to various situations to satisfy multipurpose requirements. Similarly, the setting of the center of compression can be a single point or multiple points to accommodate the requirements of the complex tensile-strain structure of other materials, such as a racket head. The material design of a racket head is required to take into account the tensile or reactive forces when stringing the racket or hitting a shuttlecock. The compression experimental simulation is then performed using explicit dynamic analysis with the finite element model. A combination of two selected material model coefficients is assigned to a single element so that it becomes a soft material and a hard material, and is further customized so that multiple soft materials and multiple hard materials are arranged in a material distribution simulator. The ratio of the soft material to the hard material changes according to the changing requirements.The material distribution simulant may be a random polygon such as a triangle, square, pentagon, or hexagon, or a random asymmetric shape, but the present invention is not limited thereto. The asymmetric shape may be a planar shape such as the sole of a foot or the palm of a hand, or it may be a solid article such as a racket, bat, golf club, engine bearing, piston, tire, tire frame, hydraulic valve body, planar shell, or planar wing, but the present invention is not limited thereto. Subsequently, after the customization of the material distribution simulant is complete, the ambient conditions and predetermined displacement control are also customized. The material distribution simulant is stationary, and the rigid compressor moves downward, compressing the material distribution simulant for a specific compression time to export the material simulation parameters, including strain energy (SE), reaction force (RF), or mean von Mises stress (ST). The speed at which the rigid compressor moves downward for compression may be constant velocity, constant acceleration, or unequal acceleration. Incidentally, the purpose of fixing the ratio of soft and hard materials is to optimize the arrangement of materials while minimizing material consumption. To save computation time for the simulation, in a preferred embodiment, the compression experiment simulation is performed by customizing the symmetric simulation.

[0045] The following disclosure illustrates a specific embodiment of the bio-inspired smart material design platform (100). Figure 2A is a flowchart illustrating the basic operating concept of the platform (100). Firstly, as shown in step A1, a compression experiment is performed by the experimental structure generator (41) on a programmatic basis to construct a 3D model of the experimental structure, and the experimental structure is manufactured by a 3D printer for compression experiments by the structure compressor (42). Secondly, as shown in step A2, a reduced model is constructed based on the compression data obtained from the compression experiment. The compression data is post-processed by the master curve generator (31) to generate a master curve corresponding to all of the compression data. The master curve referred to herein is the correlation between the relative density and intensity of the experimental structure. Next, the material coefficient curve fitting simulator (32) reads the master curve for curve fitting of the experimental structure to obtain a combination of two material coefficients, and the target material generator (33) assigns the combination of material coefficients to a single voxel as a soft material and a hard material so that the voxel has the characteristics of a composite structure. Thirdly, as shown in step A3, the finite element simulator (11) performs a simulated material distribution. The finite element simulator (11) is used to simulate the compression process of the target material to obtain a strain distribution. Fourthly, as shown in step A4, reinforcement learning is performed by the deep learning framework (21) running a reinforcement model. The material simulation parameters obtained from the material distribution simulation are used by the deep learning framework (21) to match the material distribution to the requirements of the target design. The deep learning framework (21) runs the reward function model (22) to calculate the reward value of the corresponding target material distribution. The material design that matches the target material distribution is printed using a 3D printer for experimental testing.

[0046] In one exemplary embodiment, as shown in Figure 3, unit microstructures corresponding to each shape are generated based on the formula for triple periodic minimum surfaces (TPMS), and experimental subjects with dimensions of 50 × 50 × 25 mm in length, width, and height are printed using a 3D printer. Subsequently, compression experiments are performed using a universal testing machine.

[0047] In an exemplary embodiment, gyroid, primitive, F-RD, Fischer-Koch S, diamond, or I-WP is selected as the unit microstructure of the experimental subject. For the compression experiment, three different relative densities corresponding to each unit microstructure are selected. Specifically, the experimental subject is placed flat on the compression platform of the universal testing machine. The height of the experimental subject is compressed to 75% of its initial height. In practice, an experimental subject with a height of 25 mm is compressed to a height of 18.75 mm. The compression rate is customized to 1.5 mm per minute, and the experimental process is recorded as a video. The compression experiment makes it possible to obtain a stress-strain curve corresponding to each experimental subject. Figure 4 shows the stress-strain curve of the experimental subject in the exemplary embodiment. The compression data for each experimental subject is obtained from the stress-strain curve. The compression data includes Young's modulus, plateau stress, and relative density, the relative density being calculated based on the following formula (1). JPEG2026527427000003.jpg16158

[0048] After obtaining the compression data of the subjects in these experiments, the behavior curves of the compressed porous materials can be further inferred from Ashby et al.'s work ("Lorna J. Gibson, Michael F. Ashby, "Cellular Solids: Structure and Properties, 1988"), and master curves for each experimental subject can be obtained. The properties of structural materials with various relative densities can be obtained from these master curves. As the relative density changes, the experimental subjects will have structures with various wall thicknesses. Because there are so many types of structures, it is difficult to print all of them for experimentation. This invention significantly saves experimental time by determining the mechanical properties of the structural material from the master curves.

[0049] In one exemplary embodiment, as shown in Figure 5A, the relationship diagram corresponding to the relative density and Young's modulus of the subject in each experiment is obtained by equation (2). Figure 5B shows the relationship diagram established by equation (3), where each curve corresponds to the relative density and plateau stress of the subject in each experiment. JPEG2026527427000004.jpg31134

[0050] The meanings of the symbols in equations (2) and (3) are as follows: E* is the Young's modulus of the subject in the experiment, and E S ρ is the Young's modulus of the test material in the experiment, and ρ* is the relative density of the test material in the experiment. S σ is the relative density of the test material in the above experiment, pl * represents the stress on the subject in the experiment. Substitute the parameters for each experiment into the formula and obtain the logarithm. The trend of the curve and the position of the subject in each experiment on the master curve can be observed, and here, E S This is the Young's modulus of the experimental subject, obtained by dividing by the Young's modulus of the polymer resin, in order to purely calculate the strength effect of the structure. This determines the strength of the corresponding structural material by its relative density, regardless of the type of material used.

[0051] As shown in Figure 6, in order to fit material coefficients, the experimental structure is selected, and then the corresponding material coefficients can be obtained from the master curve. It should be noted that the experimental structure for material coefficient fitting may be two or more identical or different experimental structures. Specific material coefficients such as Young's modulus, plateau stress, and relative density are obtained and then substituted into the formula disclosed by Prager (1941). The stress-strain curve of the experimental subject is simplified to an ideal stress-strain curve, from which some data are randomly selected and fed into a material coefficient curve fitting simulator (32) to perform parameter fitting. In this example, ABAQUS is used for meter fitting, and combinations of material model coefficients for the hyperelastic or Ogden model are obtained, and these combinations of material model coefficients represent the deformation behavior and strength of the selected experimental structure in the master curve.

[0052] Next, the target material generator (33) assigns the combination of material model coefficients to a single element, which has the mechanical properties of the corresponding experimental structure, and this is referred to as a reduced model (RM). After obtaining these reduced models, which are the target materials necessary for subsequent bio-inspired material design, the ABAQUS model of the finite element simulator (11) can construct two parts. The first part is the material distribution simulant composed of the target material. In this example, the material distribution simulant is a square plate that is 1 / 4 of a total of 16 single elements, divided into 4x4 in length and width. The second part is a rigid compression. In this example, the rigid compression is a rigid hemispherical plate.

[0053] Furthermore, the center point of the square plate is set as a reference point, and ABAQUS is used to perform a compression test simulation. Two combinations of material model coefficients are selected and assigned to the single element as soft material and hard material, respectively. In the 16 soft and hard materials of the plate, the ratio of soft material to hard material is 3:1. Next, boundary conditions are set by default displacement control, and the square plate is set symmetrically along the x-axis and y-axis, and the bottom surface of the square plate is fixed. The rigid hemispherical plate compresses the square plate downward at a speed of 0.5 mm / s, and the total compression time is 1 second. It should be noted that the material distribution is optimized within limited material consumption by setting the square plate to 1 / 4 the size for the symmetric simulation. Since the compression point of the rigid hemispherical plate is located at the center of the square plate, and the material design space is reduced to 1 / 4 by the symmetric simulation, the calculation time can be significantly reduced. Figure 7 shows the material simulant compression process using the finite element simulator (11). In the grid setting, the grid shape of the square plate is hexagonal, however, the present invention is not limited to this. The grid shape may be any polygon, such as triangles, rectangles, pentagons, or lattice units of any metastructure. Also, the grid shape of the rigid hemispherical plate is quad-dominated, however, the present invention is not limited to this. In this example, the center of compression of the square plate is set up with elements for analyzing the mean stress of the compressed center.

[0054] After the compression simulation of the material distribution simulant by the finite element simulator (11), the combination of material model parameters can be exported as an evaluation criterion for the subsequent reinforcement learning module (2) to perform deep learning. The user can select different combinations of material model parameters to construct the optimal target parameters (P) based on the material design requirements. For example, the combination of material model parameters may include material parameters such as strain energy (SE), reaction force (RF), mean von Mises stress (ST), and relative density (RD). In this example, the strain energy SE represents the amount of energy absorbed by the material distribution simulant, with a higher value of strain energy SE indicating higher energy absorption capacity. The reaction force RF represents the negative acceleration value reflected by post-processing in the impact test of the material distribution simulant. A smaller value of reaction force RF results in a smaller negative acceleration value. The mean von Mises stress ST represents the degree of stress concentration in the material distribution simulant during the compression simulation. The relative density RD is a density parameter obtained by dividing the target material constituting the material distribution simulator by the polymer resin. The lower the value of the relative density RD, the lower the corresponding material cost.

[0055] The simulation compression employed in this embodiment is understood to be performed by displacement control of a spherical rigid plate. Under the same displacement conditions, the greater the force, the greater the strain energy produced. Similarly, the force generated by a 5% compression of the rigid material is greater than that of the soft material, and the strain energy attributable to the rigid material is also greater. Furthermore, by performing the simulation compression using a spherical rigid plate, material simulation parameters such as strain energy and reaction force are exported, which increase as the compression displacement near the center of the compression increases.

[0056] In various embodiments, the bio-inspired smart material design platform (100) calculates a reward value (Q) corresponding to the optimal target parameter (P) by the reward function model (22) to evaluate whether the simulated material distribution results satisfy the optimal target parameter (P) customized by the user. Specifically, the reward function model (22) includes using the deep learning framework (21) to convert an initial distribution simulator to the iterative material simulator to obtain a variable quantity (D), assigning an initial reward value (q1) to the initial distribution simulator and an ending reward value (q2) to the iterative material simulator to calculate a standardized reward value (Q), and the deep learning framework (21) evaluating whether the iterative material simulator satisfies the optimal target parameter (P) based on the standardized reward value (Q).

[0057] In a preferred embodiment, the standardized reward value (Q) is calculated based on the following formula (4). Q = [q1q2] * d * P ... (4) In the formula, the standardized coefficient is calculated based on formula (5) below. d = σ[(D-α) / β]……(5) In the formula, α is the mean value of the variable (D), β is the standard deviation of the variable (D), and σ is the sigmoid function. Incidentally, d as a standardization coefficient is used to standardize the values ​​of multiple design targets in order to set a comparison range for each target. Such standardization prevents the reward value calculation in the backend from becoming biased towards a particular target design, resulting in the inability to satisfy the optimal target parameters.

[0058] In the embodiments described above, the initial distribution simulant is generated by the finite element simulator (11), which includes arranging the target material to generate the initial distribution simulant based on the optimal target parameter (P).

[0059] In the embodiments described above, the deep learning framework mimics the operating mechanism of human neural networks. The deep learning framework includes, but is not limited to, multilayer perceptrons, deep neural networks (DNNs), convolutional neural networks (CNNs), or recurrent neural networks (RNNs). Preferably, the deep learning framework is Deep Q-Networks.

[0060] The operating principle of the reinforcement learning module (2) is illustrated below. Taking Deep Q-Networks as an example, as shown in Figure 8, the biologically inspired material design is performed by the reinforcement learning model based on the material simulation parameters obtained in the above example. After the user sets the optimal target parameter (P), ABAQUS generates the initial distribution simulator by placing the target material using a scaled-down model. In this example, in the initial state of reinforcement learning, multiple initial distribution simulators are represented separately as DQN1, DQN2, DQN3, etc. In the subsequent reinforcement learning process, ABAQUS continuously changes the positions of the soft and hard materials and performs iteratively updated material distribution simulations. The iterative material simulator comprises multiple iterative distribution simulators, and in each iteration, a distribution simulator is generated consisting of target materials having a different arrangement and distribution pattern than the other simulators. Next, the Deep Q-Networks decide what action to take on these distribution simulations, and the computation process is multiplied by a decision value (D) that represents the amount of change generated between the preceding distribution simulant and the next distribution simulant. Then, the reinforcement learning module (2) assigns a reward value (Q) to each distribution simulant.

[0061] Since the setting of the optimal target parameters (P) includes various material simulation parameters, each distribution simulation is performed based on multiple material simulation parameters. To equally compare the distribution simulations corresponding to each time point in the distribution simulation, the reward value (Q) is standardized by a coefficient d, and the standardized reward value (Q) is calculated to determine which distribution simulants have a greater impact on the design. Furthermore, whether the final iterative material simulant fits the setting of the optimal target parameters (P) is evaluated based on the standardized reward value (Q). In this example, referring again to Figure 8, α and β are the mean and standard deviation of the variable quantity (D), respectively, and σ represents the sigmoid function. During the training process, distribution simulations with high standardized reward values ​​(Q) are recorded, and the distribution simulation with the best standardized reward value (Q) is selected and summed. Subsequently, the distribution simulation is iteratively performed to obtain the final target material distribution, which is the material distribution design that fits the setting of the optimal target parameters (P).

[0062] Furthermore, the bio-inspired smart materials design platform (100) further comprises a materials design module (5) that communicates with the reinforcement learning module (2) to design a bio-inspired simulated distribution simulant based on a second optimal target parameter customized by the user. The second optimal target parameter is customized by the materials simulation parameters, including strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more of these.

[0063] In other areas, as shown in Figure 2B, the present invention provides a method for designing bio-inspired smart materials, the method comprising the steps described below. In step S1, a target material distribution simulation is performed by a material distribution simulation module (1) to obtain material simulation parameters. The target materials include rigid and flexible materials set by different combinations of material model coefficients. In step S2, a reward function model (22) is executed in a deep learning framework (21) configured as a reinforcement learning module (2) for calculating reward values ​​and evaluating whether iterative material simulants fit the optimal target parameters (P). The reward values ​​are calculated to correspond to iterative material simulants of other target materials based on the optimal target parameters (P). The optimal target parameters are set based on the material simulation parameters, including strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more of these.

[0064] In one or more embodiments, the reward calculation includes: changing the initial distribution simulator in the deep learning framework (21) to the iterative material simulator and obtaining a variable quantity (D); assigning an initial reward value (q1) to the initial distribution simulator and assigning a final reward value (q2) to the iterative material simulator and calculating a standardized reward value (Q); and evaluating in the deep learning framework, based on the standardized reward value (Q), whether the iterative material simulator fits the optimal target parameter (P).

[0065] In a preferred embodiment, the standardized reward value (Q) is calculated based on the following formula (4). Q = [q1q2] * d * P ... (4) In the formula, d is the standardized coefficient calculated based on the following formula (5). d = σ[(D-α) / β]……(5) In the formula, α is the mean value of the variable (D), β is the standard deviation of the variable (D), and σ is the sigmoid function.

[0066] In one or more embodiments, the material distribution simulation includes exporting the material simulation parameters by generating a material distribution simulant for the target material using a finite element simulator (11), generating a stiff compressible body, and simulating the stiff compressible body to compress the material distribution simulant.

[0067] Preferably, the initial distribution simulator is generated by the finite element simulator (11) based on the optimal target parameter (P).

[0068] In a preferred embodiment, referring to Figure 2B, the method further comprises a reduced model configuration. In step S3, the reduced model configuration comprises: generating a master curve by a master curve generator (31) based on compressed data, wherein the compressed data includes Young's modulus, plateau stress, and relative density; performing a curve fitting of a selected experimental structure to the master curve by a material coefficient curve fitting simulator (32) to obtain a combination of material coefficients and export the combination of material model coefficients; and assigning the combination of material model coefficients to a single element by a target material generator (33) to obtain the target material, wherein the combination of material model coefficients includes a combination of hard material model coefficients, a combination of soft material model coefficients, or a combination thereof.

[0069] Preferably, the experimental structure includes a gyroid, primitive, F-RD, Fischer-Koch S, diamond, or I-WP. The experimental structure is generated based on a metastructure model.

[0070] In a preferred embodiment, referring to Figure 2B, in step S3” the acquisition of the compressed data includes generating the compressed data by the compression experiment module (4). A metastructure is generated by the experimental structure generator (41) based on a metastructure model, and the metastructure is compressed by the structure compressor (42) to acquire the compressed data. The metastructure is either the same as or different from the experimental structure.

[0071] Preferably, the experimental structure comprises a gyroid, primitive, F-RD, Fischer-Koch S, diamond, or I-WP, and the experimental structure is generated based on a metastructure model.

[0072] In more detail, the aforementioned metastructure model includes a triple periodic minimal surface model (TPMS model).

[0073] Preferably, referring to Figure 2B, as shown in step S4, the method further includes designing a bio-inspired simulated distribution simulant by the material design module (5) based on a second optimal target parameter customized by the user. The second optimal target parameter is customized by the material simulation parameters, including strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more thereof.

[0074] To specifically demonstrate that the biologically inspired smart material design platform (100) and method according to the present invention realize highly efficient and highly accurate goal-oriented material design, the technical effects of the platform (100) are further verified by the following embodiments.

[0075] Example 1: Shoe Midsole Design The design of the shoe's midsole requires at least three structures with different objectives. Figures 9A to 9C show three material distribution simulants and their corresponding simulated strain distribution diagrams. These material distribution simulants were customized by the bio-inspired smart material design platform (100) based on three different sets of optimal objective parameters.

[0076] Firstly, as shown in Figure 9A, the material distribution simulator 1 is related to only one target function, and the material distribution simulator 1 is intended to design a shoe midsole for an arch, with a design space of 8x8. The target structure was expected to provide high load-bearing capacity. For this reason, the optimal target parameters were customized to maximize the reaction force (RF). Reinforcement learning calculations showed that the rigid material was concentrated at the center of the compression, improving the load-bearing capacity.

[0077] Secondly, as shown in Figure 9B, the material distribution simulator 2 has the same design space as the material distribution simulator 1, but the material distribution simulator 2 considers designs for multiple targets, and was intended to design a midsole for the heel. The target structure was expected to absorb more strain energy. By reducing the excess energy generated by the bottom of the heel and transmitted to the user's foot, discomfort felt during prolonged walking can be prevented. Furthermore, by preventing excessive rebound force on the structure due to excessively concentrated stress, the rebound force from the heel is not transmitted to joints such as the knee. In this way, the optimal target parameters were customized to maximize the strain energy (SE) in the central part of the shoe's midsole and minimize the reaction force (RF) and mean von Mises stress (ST).

[0078] Thirdly, as shown in Figure 9C, the material distribution simulator 3 had a 12x12 identical design and was designed to fit multiple targets. The material distribution simulator 3 was intended to design a shoe midsole for the toe of the foot and was expected to have a softer target structure. Therefore, it was necessary to reduce stress concentration and increase strain energy absorption in order to protect the structurally fragile toe of the foot. Accordingly, the optimal target parameters were customized to maximize the strain energy (SE) in the central portion of the shoe midsole and minimize the mean von Mises stress (ST).

[0079] Table 1 lists the customized optimal target parameters for the three different material distribution simulators mentioned above. Table 1 JPEG2026527427000005.jpg35170

[0080] Table 2 lists the simulated material parameters for the three different material distribution simulators described above, which were simulated using the platform (100). Table 2 JPEG2026527427000006.jpg35152

[0081] After the optimal material distributions for the three different material distribution simulants were determined, the experimental structure generator (41) program created 3D-printed models for each simulant. These three designs were then manufactured using a 3D printer and named specimen 1 (simulant 1), specimen 2 (simulant 2), and specimen 3 (simulant 3), respectively. In actual testing, to verify whether the aforementioned designs met the optimal target parameters, a TM-142 test machine was used for the drop weight impact absorption test in this example. The test parameters included maximum deceleration, rebound value, maximum indentation depth, original thickness, and thickness after impact.

[0082] As shown in Table 3, the main criteria were the maximum deceleration (g) and the rebound value. The smaller the g value and the rebound value, the greater the energy that the designed bio-inspired material structure could absorb, and the larger the g value, the greater the rebound energy. When each of the manufactured test specimens was examined, the g value of test specimen 1 was 8 and the rebound value was 2%, the g value of test specimen 2 was 7 and the rebound value was 2%, and the g value of test specimen 3 was 9 and the rebound value was 4%. The values ​​for the three energy absorption performances were lower than those obtained in conventional impact absorption tests. For example, the g value in conventional impact tests is usually in the range of 9 to 15. This indicates that the bio-inspired material structure designed by the platform (100) can achieve the objective of absorbing more energy, regardless of whether a single target parameter is set or multiple target parameters are set, and that it satisfies the customization of the optimal target parameters.

[0083] Table 3 JPEG2026527427000007.jpg48170

[0084] In addition, test specimen 1, which was designed for a single target, was expected to have higher load-bearing capacity due to the concentration of the hard material in the center of the shoe's midsole. As shown in Table 3, the thickness of test specimen 1 after impact was 13.9 mm, which was thicker than both test specimens 2 and 3 (7.0 mm and 10.2 mm, respectively). Clearly, test specimen 1 showed superior load-bearing capacity compared to both test specimens 2 and 3, and its thickness after impact was thicker than the other two. On the one hand, test specimen 1 had high load-bearing capacity, but in the impact test, test specimen 1 generated the largest maximum indentation depth (9.5 mm), which was deeper than the other two (4.5 mm and 5.0 mm, respectively). When subjected to the impact of the weight, test specimen 1 generated greater strain energy under stress than the other two test specimens, which was consistent with the target setting of ignoring strain energy (SE) when customizing the parameters. In customizing the parameters, when strain energy (SE) was included in the parameter settings for test specimen 2 and test specimen 3, less strain energy was generated during the impact test, which also conformed to the expected target setting.

[0085] The bio-inspired smart material design platform according to the present invention, when combined with a deep learning framework and a standard set of design processes, can simulate the corresponding material very effectively, provided that the reward function model of the reinforcement learning model is clearly defined. The material distribution design can be adapted to the material design requirements in a variety of situations and is not limited to the structure of the article to be simulated. The arrangement and distribution of materials can be designed to match the structure of the article, such as planar, three-dimensional, symmetrical, or asymmetrical, and has a wide range of potential applications.

[0086] The present invention utilizes the scaled-down model configuration. By assigning the microstructure coefficients to the target material, specific mechanical properties are imparted to the target material. In the material design, including composite structures, the scaled-down model can be used in subsequent symmetric or asymmetric simulation methods. The fixed amount of target material significantly reduces simulation computation time, thereby lowering the time cost of material design.

[0087] The bio-inspired smart material platform according to the present invention specifically performs simulated design of bio-inspired materials by combining bio-inspired material structures, finite element simulations, and a deep learning framework. As shown in Figure 10, the potential application level of the present invention is very broad. As long as material parameters of bio-inspired material structures, such as compression, tension, extension, buckling, or other parameters, are available, the user can customize the optimal target parameters according to the demands of the backend product, and further use the platform to simulate bio-inspired material structure distributions that fit the optimal target parameters, and then the product is manufactured based on the simulation results.

[0088] The biologically inspired smart material design platform according to the present invention effectively reduces reaction forces and stress concentration phenomena when a product is subjected to impact through parameter fitting and model construction, which is beneficial for the material design of products with complex microstructures. As shown in Figure 10, the platform is expected to be widely used in various industrial fields such as shoe sole design, one-piece molded sandals, badminton racket grip design, helmet design, golf structural design, military defense, aerospace, automotive industry, or other industrial fields.

Claims

1. A material distribution simulation module configured to perform a target material distribution simulation in order to obtain material simulation parameters, The system comprises a reinforcement learning module configured with a deep learning framework that executes a reward function model to communicate with the material distribution simulation module, calculate the reward value of an iterative material simulator of other target materials based on the optimal target parameter (P), and evaluate whether the iterative material simulator fits the optimal target parameter (P), Here, the target materials include hard materials and soft materials, and each material is assigned a different combination of material model coefficients. A bio-inspired smart materials design platform characterized in that the optimal target parameters are set based on material simulation parameters including strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more thereof.

2. The aforementioned reward function model is: The initial distribution simulator is modified to the iterative material simulator in order to obtain the variable quantity (D) based on the deep learning framework, To calculate the standardized reward value (Q), an initial reward value (q1) is assigned to the initial distribution simulator, and a final reward value (q2) is assigned to the repeating material simulator. The bio-inspired smart materials design platform according to claim 1, characterized in that it includes evaluating whether the iterative material simulator fits the optimal target parameter (P) based on the standardized reward value (Q) using the deep learning framework.

3. The bio-inspired smart materials design platform according to claim 2, characterized in that the standardized reward value (Q) is calculated based on the following formula. Q=[q1q2]*d*P, (In the formula, d is the standardized coefficient calculated based on the formula below.) d=σ[(D-α) / β], (In the formula, α is the mean value of the variable (D), β is the standard deviation of the variable (D), and σ is the sigmoid function.)

4. The bio-inspired smart material design platform according to claim 1, characterized in that the material simulation parameters further include stiffness modulus, elastic modulus, shear modulus, strain stress, deformation modulus, or strain variation coefficient.

5. The aforementioned material distribution simulation module is: The bio-inspired smart material design platform according to claim 1, comprising a finite element simulator configured to generate a material distribution simulant using the target material, generate a rigid compressible body, and export the material simulation parameters by simulating the rigid compressible body to compress the material distribution simulant.

6. To construct the target material, a scaled-down model configuration module is provided that communicates with the material distribution simulation module, and the scaled-down model configuration module is A master curve is generated based on compressed data, and the compressed data includes a master curve generator containing Young's modulus, plateau stress, and relative density. A material coefficient curve fitting simulator performs curve fitting of the selected experimental structure to the aforementioned master curve, obtains combinations of material coefficients, and outputs combinations of material model coefficients. The bio-inspired smart material design platform according to claim 1, comprising a target material generator that assigns the combination of material coefficients to a single element to obtain the target material, wherein the combination of material model coefficients includes a combination of hard material model coefficients, a combination of soft material model coefficients, or a combination thereof.

7. The bio-inspired smart materials design platform according to claim 6, characterized in that the experimental structure includes gyroid, primitive, F-RD, Fischer-Koch S, diamond, or I-WP, or the experimental structure is generated based on a metastructure model, the metastructure model includes a triple periodic minimal surface model (TPMS model).

8. The system further comprises a compression experiment module connected to the aforementioned reduced model configuration module, The bio-inspired smart material design platform according to claim 1, wherein the compression experiment module includes an experimental structure generator for generating a metastructure based on a metastructure model, and a structure compressor for compressing the metastructure to obtain the compression data, wherein the metastructure is the same as or different from the experimental structure.

9. The bio-inspired smart materials design platform according to claim 1, further comprising a materials design module that communicates with the reinforcement learning module to design a bio-inspired simulated distribution based on a second optimal target parameter customized by the user, wherein the second optimal target parameter is set by reference to the materials simulation parameters, which include strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more thereof.

10. To obtain material simulation parameters, the material distribution simulation module performs a target material distribution simulation, where the target material includes hard and soft materials, and each material is assigned a different combination of material model coefficients. A bio-inspired smart material design method comprising: running a reward function model using a deep learning framework configured as a reinforcement learning module for calculating reward values ​​and evaluating whether the iterative material simulant fits the optimal target parameters (P); the reward values ​​are calculated to correspond to iterative material simulants of other target materials based on the optimal target parameters (P), the optimal target parameters being set by reference to material simulation parameters including strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more thereof.

11. The bio-inspired smart material design method according to claim 10, characterized in that the material simulation parameters further include stiffness modulus, elastic modulus, shear modulus, strain stress, deformation modulus, or strain variation coefficient.

12. The calculation of the aforementioned reward value is as follows: The reward function model modifies the initial distribution simulator to the iterative material simulator in order to obtain the variable quantity (D) based on the deep learning framework, To calculate the standardized reward value (Q), an initial reward value (q1) is assigned to the initial distribution simulator, and a final reward value (q2) is assigned to the repeating material simulator. A method for designing bio-inspired smart materials according to claim 10, comprising: evaluating whether the iterative material simulator fits the optimal target parameter (P) based on the standardized reward value (Q) using the deep learning framework.

13. The method for designing bio-inspired smart materials according to claim 12, characterized in that the standardized reward value (Q) is calculated based on the following formula. Q=[q1q2]*d*P (In the formula, d is the standardized coefficient calculated based on the formula below.) d=σ[(D-α) / β], (In the formula, α is the mean value of the variable (D), β is the standard deviation of the variable (D), and σ is the sigmoid function.)

14. The simulation of the aforementioned material distribution is performed as follows: A bio-inspired smart material design method according to claim 10, characterized in that it includes generating a material distribution simulant of the target material using a finite element simulator, generating a rigid compressible body, and exporting the material simulation parameters by simulating the rigid compressible body that compresses the material distribution simulant.

15. The bio-inspired smart material design method according to claim 10, characterized in that the material distribution simulation module comprises a finite element simulator configured to generate the initial distribution simulator by arranging the target material based on the optimal target parameter (P).

16. The scaled-down model configuration is provided, and the scaled-down model configuration is A master curve is generated by a master curve generator based on compressed data, and the compressed data includes Young's modulus, plateau stress, and relative density. The material coefficient curve fitting simulator is used to obtain combinations of material coefficients and export combinations of material model coefficients, thereby performing curve fitting of the selected experimental structure to the master curve. A bio-inspired smart material design method according to claim 10, characterized in that a target material generator assigns a combination of material coefficients to a single element in order to obtain the target material, wherein the combination of material model coefficients includes a combination of hard material model coefficients, a combination of soft material model coefficients, or a combination thereof.

17. The method for designing bio-inspired smart materials according to claim 16, characterized in that the experimental structure includes gyroid, primitive, F-RD, Fischer-Koch S, diamond, or I-WP, or the experimental structure is generated based on a metastructure model, the metastructure model includes a triple periodic minimal surface model (TPMS model).

18. The method for designing bio-inspired smart materials according to claim 10, characterized in that the acquisition of the compressed data includes generating the compressed data using a compression experiment module.

19. The generation of the aforementioned compressed data is The experimental structure generator generates metastructures based on metastructure models, and A method for designing a bio-inspired smart material according to claim 18, comprising compressing the metastructure with a structural compressor so as to obtain the aforementioned compressed data, wherein the metastructure is the same as or different from the aforementioned experimental structure.

20. The bio-inspired smart material design method according to claim 10, further comprising designing a bio-inspired simulated distribution simulator by a material design module based on a second optimal target parameter customized by the user, wherein the second optimal target parameter is set with reference to the material simulation parameters, which include strain energy (SE), reaction force (RF), mean von Mises stress (ST), or a combination of two or more thereof.

Images