An intelligent design method for multi-cellular structures with concave-convex minimal surfaces and controllable dynamic mechanical response curves

Abstract

The present invention provides an intelligent design method for multicellular structures with concave-convex wall surfaces and controllable dynamic mechanical response curves. The method changes the topological geometric parameters and concave-convex characteristic parameters of the unit cell structures of the concave-convex wall surfaces, combines the unit cell structures to form multicellular structures, and combines the machine learning model to realize the intelligent design of the multicellular structure. First, a parameterized minimal unit cell structure with concave-convex wall surfaces is designed, and then a series of unit cell structures with different parameters are combined to form a unit cell matrix. The test pieces are prepared by additive manufacturing methods and experimental tests are carried out. At the same time, loading numerical simulations are carried out, and dynamic response results are obtained through experiments and simulations. The combined unit cell structure is constructed to form a multicellular structure training data set, and the machine learning model is trained. Finally, the unit cell and multicellular structures with concave-convex wall surfaces that meet the requirements are predicted and prepared according to the actual engineering needs. This method realizes the regulation of the structural bearing capacity and meets the needs of different applications.

Description

Technical Field

[0001] The present invention relates to the field of material structure technology, and in particular to an intelligent design method for a multi-cellular structure with concave-convex wall minimal curved surface capable of regulating a dynamic mechanical response curve. Background Art

[0002] In recent years, the study of cellular structures and the mechanical properties of their materials has been one of the research hotspots in the field of mechanics. Due to their advantages such as light weight and high strength, strong impact resistance and energy absorption, shock absorption and noise reduction, they have huge application potential in aerospace, transportation, national defense science and technology, biomedicine, energy, mechanical equipment and other fields.

[0003] As a typical representative of these, the three-periodic minimal surface (TPMS) polyhedral structure has attracted much attention due to its superior performance such as high lightness and low weight, and has a strong prospect for results transformation and application breadth. For this reason, various scientific research institutions have carried out a lot of research work on these basic topological configurations of TPMS polyhedral structures, involving processing and preparation, quasi-static mechanical properties, dynamic mechanical properties and other aspects, and explored the relationship between structural characteristic parameters such as relative density and pore size and mechanical properties.

[0004] To meet complex practical needs, researchers have also tried to expand the above-mentioned basic TPMS topological configuration to TPMS multicellular structures, TPMS mixed multicellular structures, and TPMS fractal nested multicellular structures, and designed, prepared, and analyzed the dynamic mechanical properties of these structures.

[0005] But in fact, the mechanical properties of cellular materials are closely related to their microstructures. Therefore, the mechanical properties of cellular materials can be controlled by adjusting their microconfiguration. In addition, cellular structures will have significant differences due to different conditional parameters. In order to explore the influence of these microconfigurations and conditional parameters on the mechanical properties of cellular mechanisms, it is necessary to prepare various cellular mechanisms under different microstructures and different conditional parameters into samples, and study the cellular structures through experimental simulation analysis methods. This experimental method is obviously cumbersome, repetitive and costly.

[0006] In order to solve the above problems, this scheme provides an intelligent design method for multi-cellular structures with concave-convex walls and minimal curved surfaces that can control the dynamic mechanical response curve. Summary of the Invention

[0007] In order to solve the above technical problems, an embodiment of the present invention provides an intelligent design method for a multi-cellular structure with concave-convex minimal surfaces on the wall, which can control the dynamic mechanical response curve, comprising the following steps:

[0008] Step S1: designing a wall concave-convex parameterized minimal surface unit cell structure based on a three-periodic minimal surface unit cell structure;

[0009] Step S2: by changing the topological geometric parameters and concave-convex characteristic parameters of the unit cell structure of the concave-convex minimal surface on the wall, unit cell structures with different topological geometric parameters and concave-convex characteristics are formed, and a unit cell matrix is constructed;

[0010] Step S3: using an additive manufacturing method to prepare a test piece of a minimal surface unit cell structure with a concave-convex wall surface, and conducting experimental tests under experimental conditions of different loading rates;

[0011] Step S4: performing loading numerical simulation on the concave-convex minimal surface unit cell structure with topological geometric parameters and concave-convex features obtained in step S2;

[0012] Step S5: Based on the dynamic response results of the unit cell structure obtained in steps S3 and S4, a parameter matrix of topological geometric parameters and unit cell concavity and convexity and a unit cell mechanical response matrix are constructed to form a training data set and a machine learning model is constructed for training, so that the mechanical training model can predict the topological geometric parameters and unit cell concavity and convexity parameters of the unit cell structure of the minimal surface with concave and convex wall according to the mechanical response form;

[0013] Step S6: combining the unit cell structures to form an N*N*N polycellular structure, performing experiments and simulations on the polycellular structure of the combined unit cell structure to obtain polycellular structure response data, performing machine learning classification and training using the unit cell structure combination as an input parameter set and the polycellular structure response data as a target set, to obtain a polycellular structure machine learning model with concave-convex minimal curved surfaces on the wall that can control the dynamic mechanical response curve;

[0014] Step S7: Based on the machine learning prediction results described in steps S5 and S6, the parameters of the single-cell structure and multi-cell structure of the concave-convex minimal surface on the wall that meet the actual engineering mechanics response target are obtained and the preparation is completed. The topological geometric parameters of the concave-convex minimal surface structure on the wall in step S2 are determined by the C value in the generalized governing equation of the three-periodic minimal surface: Cos(x)+Cos(y)+Cos(z)=C, and the value range of the C value is -1≤C≤1.

[0015] The concave-convex characteristic parameters of the minimal curved surface unit cell structure with concave-convex wall surface in step S2 include: concave-convex size, concave-convex depth and concave-convex distribution.

[0016] The additive manufacturing method in step S3 includes but is not limited to any one of photocuring, melt extrusion, and selective laser melting.

[0017] The loading rate of the dynamic mechanical loading experiment in step S3 may be 0.01 m / min to 5 m / min.

[0018] The dynamic response results in step S5 should at least include total absorbed energy, initial peak force, average crushing force, and stress-strain curve.

[0019] The combination of the unit cell structures in step S6 to form the N*N*N multi-cell structure is not limited to being uniform, interlayer gradient, or orthogonal anisotropy.

[0020] The mechanical response target required for the actual project in step S7 may be one or more of the targets such as energy absorption, peak force, and average crushing force.

[0021] Implementation of the embodiments of the present invention has the following beneficial effects: ① The intelligent design method for multicellular structures with concave-convex minimal surfaces and controllable dynamic mechanical response curves provided by the present invention can predict and prepare single and multicellular structures with concave-convex minimal surfaces that meet actual engineering requirements by varying topological geometric parameters and concave-convex parameters. This method enables the regulation of structural load-bearing capacity to meet diverse application requirements.

[0022] ② The present invention provides a method for designing polycellular structures, which changes the microstructure of a single-cell structure, and combines single-cell structures to form a polycellular structure in conjunction with simulation and experiments. Combined with a machine learning model, the machine learning model can be trained to predict various mechanical parameters of the polycellular structure, and the design of the polycellular structure is guided by the predicted data. The use of this method can greatly reduce the process of repeated preparation experiments in the laboratory's research on polycellular structures, reduce experimental costs, and improve experimental efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a specific flow chart of the present invention;

[0024] Figure 2 It is a schematic diagram of the unit cell structure of the three-periodic minimal surface Schwarz P surface and the concave-convex minimal surface on the wall;

[0025] Figure 3 This is a diagram of the process of creating a minimal surface unit cell structure with concave and convex walls;

[0026] Figure 4 This is a schematic diagram of a series of concave-convex minimal surface unit cell structures obtained by varying parameters;

[0027] Figure 5 is a schematic diagram of the finite element loading model;

[0028] Figure 6 It is a diagram of the neural network structure;

[0029] Figure 7 This is a comparison chart between the finite element simulation results and the artificial neural network prediction results;

[0030] Figure 8 It is a schematic diagram of a multicellular structure with uniformity, interlayer gradient and orthotropic anisotropy;

[0031] Figure 9 It is a schematic diagram of a two-section multi-cellular energy-absorbing composite structure replacing the anti-collision beam. DETAILED DESCRIPTION

[0032] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be described in further detail below with reference to the accompanying drawings.

[0033] Example 1

[0034] This embodiment mainly discloses an intelligent design method for a multi-cellular structure with concave-convex minimal curved surface having a controllable dynamic mechanical response curve, which specifically includes the following steps:

[0035] Step S1: designing a wall concave-convex parameterized minimal surface unit cell structure based on a three-periodic minimal surface unit cell structure;

[0036] Step S2: by changing the topological geometric parameters and concave-convex characteristic parameters of the unit cell structure of the concave-convex minimal surface on the wall, unit cell structures with different topological geometric parameters and concave-convex characteristics are formed, and a unit cell matrix is constructed;

[0037] Step S3: using an additive manufacturing method to prepare a test piece of a minimal surface unit cell structure with a concave-convex wall surface, and conducting experimental tests under experimental conditions of different loading rates;

[0038] Step S4: performing loading numerical simulation on the concave-convex minimal surface unit cell structure with topological geometric parameters and concave-convex features obtained in step S2;

[0039] Step S5: Based on the dynamic response results of the unit cell structure obtained in steps S3 and S4, a parameter matrix of topological geometric parameters and unit cell concavity and convexity and a unit cell mechanical response matrix are constructed to form a training data set and a machine learning model is constructed for training, so that the mechanical training model can predict the topological geometric parameters and unit cell concavity and convexity parameters of the unit cell structure of the minimal surface with concave and convex wall according to the mechanical response form;

[0040] Step S6: combining the unit cell structures to form an N*N*N polycellular structure, performing experiments and simulations on the polycellular structure of the combined unit cell structure to obtain polycellular structure response data, performing machine learning classification and training using the unit cell structure combination as an input parameter set and the polycellular structure response data as a target set, to obtain a polycellular structure machine learning model with concave-convex minimal curved surfaces on the wall that can control the dynamic mechanical response curve;

[0041] Step S7: Through the machine learning prediction results described in steps S5 and S6, the parameters of the wall concave-convex minimal surface single-cell structure and multi-cell structure that meet the actual engineering mechanics response target are obtained and the preparation is completed.

[0042] Better, such as Figure 2 As shown, in step S1, different types of three-periodic minimal surfaces can be used as a basis for wall surface concave-convex transformation, and the minimal surface types include, but are not limited to, Schwarz P surfaces, Schwarz Diamond surfaces, Gyroid surfaces, and I-WP surfaces.

[0043] More preferably, in step S2, the topological geometric parameters of the concave-convex minimal surface structure of the wall are determined by the C value in the generalized control equation of the three-periodic minimal surface: Cos(x)+Cos(y)+Cos(z)=C, and the value range of the C value is -1≤C≤1.

[0044] More preferably, the concave-convex characteristic parameters of the minimal curved surface unit cell structure with concave-convex wall surface in step S2 include: concave-convex size, concave-convex depth and concave-convex distribution.

[0045] Specifically, the size of the concave and convex can be controlled by the diameter β1 of the concave and convex on the wall, and the value range can be 0.15*a≤β1≤0.25*a; the depth of the concave and convex can be controlled by the Boolean distance β2, and the value range can be 0.125*β1≤β2≤0.5*β1; the thickness is β3, and the value range can be 0.001*a<β3<0.5*a, where a is the outer size of the concave and convex unit cell on the wall.

[0046] More preferably, the additive manufacturing method in step S3 includes but is not limited to any one of photocuring, melt extrusion, and selective laser melting.

[0047] More preferably, the loading rate of the dynamic mechanical loading experiment in step S3 may be 0.01 m / min to 5 m / min.

[0048] More preferably, the dynamic response result in step S5 should at least include total absorbed energy, initial peak force, average crushing force, and stress-strain curve.

[0049] More preferably, the combination of the unit cell structures to form the N*N*N multi-cell structure in step S6 is not limited to being uniform, having an interlayer gradient, or being orthogonally anisotropic.

[0050] More preferably, the mechanical response target required for the actual project in step S7 may be one or more of the targets such as energy absorption, peak force, average crushing force, etc.

[0051] Example 2

[0052] In order to better verify the intelligent design method of multi-cellular structures with concave-convex minimal surfaces on the wall and adjustable dynamic mechanical response curve proposed in this solution, this embodiment designs a mechanical study of a multi-cellular structure for illustration.

[0053] First, based on Figure 3 A minimal curved surface structure with a concave-convex wall surface is shown, and the topological geometric parameters and concave-convex characteristic parameters of the minimal curved surface structure with a concave-convex wall surface are changed.

[0054] Specifically, in this embodiment, a, that is, the size of the concave-convex wall unit cell is 20 mm, then the ball diameter β1 is adjusted within the range of 3 mm ≤ β1 ≤ 5 mm, the Boolean distance β2 is adjusted within the range of 0.5 mm ≤ β2 ≤ 2.5 mm, and the thickness β3 is adjusted within the range of 0.3 mm ≤ β3 ≤ 1 mm, and a series of concave-convex wall minimal surface unit cell structures are obtained by batch processing, as shown in the schematic diagram of some structures. Figure 4 shown.

[0055] Then, the photocuring additive manufacturing technology is used to prepare a minimal curved surface unit cell structure with a concave-convex wall surface. The photosensitive resin material is cured layer by layer by an LED light source, and then after cleaning, drying and curing, a minimal curved surface unit cell structure specimen with a concave-convex wall surface is obtained. After that, a mechanical loading test is carried out on the minimal curved surface unit cell structure specimen with a concave-convex wall surface at a loading rate of 0.3 m / min to obtain a dynamic mechanical response.

[0056] Secondly, the created concave-convex minimal surface unit cell structure is imported into the finite element software Abaqus, the material constitutive parameters are set, the cross-section model is established, the corresponding thickness is set, and it is assigned to the unit cell model.

[0057] like Figure 5 As shown, the model is assembled, all degrees of freedom of the lower rigid plate are constrained and set to 0. The degrees of freedom of the upper rigid plate are all set to 0 except the translational degree of freedom in the loading direction. A constant compression speed is set for the upper rigid plate to simulate a quasi-static load. Considering the balance between the accuracy and time of numerical simulation solution, the geometric model is meshed, and the S3R shell element is used for mesh processing and solution. The general contact model is used to consider the friction between the internal unit cell structure of the concave and convex minimal surface of the wall and the friction between the unit cell structure of the concave and convex minimal surface of the wall and the indenter. The contact property is normal hard contact, and the penalty function method is used to consider the friction in the tangential direction. A solution task is created and the calculation is submitted.

[0058] The results of all the finite element models described above were processed to obtain dynamic mechanical responses, including total absorbed energy, initial peak force, average crushing force, and stress-strain curves. The unit cell structures of the concave-convex minimal surface were then linked to the corresponding dynamic mechanical response results to form a database that correlated structural parameters with the dynamic mechanical response results. This database was then divided into a training set and a test set, which served as the dataset for machine learning training and validation.

[0059] like Figure 6 As shown, this embodiment uses an artificial neural network as a machine learning model.

[0060] 90% of the samples in the data set are randomly selected as the training set, and the remaining 10% are used as the test set. The dynamic response results are used as the input of the data set, and the topological geometric parameters of the wall concave-convex minimal surface unit cell structure and the parameters of the unit cell concavity and convexity are used as the labels of the data set. An artificial neural network model is constructed, and the training set is used to train the above-constructed machine learning model. When the loss function reaches the set maximum value range at the same time, the model parameters are saved to obtain the trained machine learning model.

[0061] The training process of this machine learning model is as follows: put some data in the training set into the artificial neural network for training, train all the training set data in rounds to obtain the training set loss value; then put the test set data into the artificial neural network model for testing and calculate the test set loss value through the loss function; then through back propagation, continuously adjust the network parameters so that the loss values of the training set and test set reach the set maximum range at the same time, and finally save the generated trained artificial neural network model, and then evaluate the trained neural network.

[0062] In this embodiment, the finite element software Abaqus is used to model and simulate the structural parameters predicted by the neural network. The predicted structural parameter simulation loading curve L1 is compared with the loading curve L2 as the input to judge the prediction effect and generalization ability of the neural network. The comparison between the finite element simulation results and the artificial neural network prediction results is shown in the figure below. Figure 7 As shown, it can be seen that the L1 and L2 curves are roughly the same, and the prediction results of the artificial neural network have only a 3.71% error compared with the results of the finite element simulation. Therefore, it can be seen that the prediction structure of the artificial neural network has a strong reference value.

[0063] The trained artificial neural network is used to predict the unit cell structure parameters of all dynamic mechanical responses, and a parameter database of dynamic mechanical response data, topological geometric parameters and unit cell concavity and convexity is constructed based on the prediction results of the artificial neural network.

[0064] The combined unit cell structure, that is, the unit cell mechanism forms an N*N*N polycellular structure. Experiments and simulations are carried out on the polycellular structure of the combined unit cell structure to obtain the polycellular structure response data. The unit cell structure combination form is used as the input parameter set and the polycellular structure response data is used as the target set for machine learning classification and training to obtain a machine learning model of a polycellular structure with concave and convex minimal curved surface on the wall with controllable dynamic mechanical response curve.

[0065] It should be noted that if Figure 8 As shown, the combination form of the N*N*N multicellular structure formed by combining the unit cell structures can be uniform, interlayer gradient, or orthogonal anisotropic.

[0066] According to the proposed reasonable dynamic mechanical response, the topological geometric parameters and concavity and convexity parameters of the minimal surface polyhedral structure with concave and convex wall are predicted by artificial neural network, and the topological geometric parameters and concavity and convexity parameters with the minimum error are returned as the recommended polyhedral structure.

[0067] Example 3

[0068] This embodiment mainly discloses a design process of using this solution to design a polyhedral material that can protect pedestrians and vehicles.

[0069] For example, in the vehicle field, in order to protect pedestrians and the vehicle itself, it is necessary to design a multicellular material that can absorb collision energy. In the event of a collision between a car and a pedestrian, a multicellular material with two-stage energy absorption needs to be designed structurally to achieve the process of absorbing energy from small to large in different stages of the energy absorption process.

[0070] When the collision stress of the first section is designed to be small, the first section structure is completely deformed and crushed to fully absorb energy and achieve the purpose of protecting pedestrians.

[0071] In the second design phase, when a car collides with other objects and the stress is high, a certain structural bearing capacity is required to protect the vehicle from excessive deformation and harming the people inside the vehicle. At this time, the structure of the cellular material can be designed to make it exhibit a large amount of energy absorption and not prone to excessive deformation, so as to achieve the purpose of absorbing a large amount of energy.

[0072] In order to adapt to both working conditions at the same time, combined with the concave-convex minimal surface multicellular structure of the wall surface of the present invention, the single cell and multicellular structure of the first section and the second section are designed according to the dynamic response, such as Figure 9 As shown, a two-stage energy absorption model is designed to replace the front anti-collision beam of the car.

[0073] With the help of dimensional modeling software, a geometric model of a minimal surface unit cell structure with a concave and convex wall size of 20*20*20mm was constructed, and then three 60*60*1mm thin plates were built on the upper, middle and lower parts to construct an energy-absorbing structure.

[0074] The first part consists of a minimal surface unit cell structure with a concave-convex wall and a gradually slowing loading curve, which is fixed with two thin plates in the upper and middle parts using special glue. When a collision between a car and a person is inevitable, the purpose of fully protecting the pedestrian is achieved.

[0075] The second part is to design a 3*3*3 combined unit cell structure based on the above "Step 6". The combination form is uniform, and the combined unit cell structure has a wall concave-convex minimal surface multi-cellular structure with a large stress and total energy absorption and a stable loading curve. The middle and lower thin plates are consolidated with special glue to achieve the purpose of absorbing a large amount of energy and not easily causing excessive deformation when colliding with other harder objects, thereby ensuring the safety of people inside the car.

[0076] The above embodiments are intended only to illustrate the technical concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the present invention and implement it accordingly. They are not intended to limit the scope of protection of the present invention. Any modifications made within the spirit of the main technical solution of the present invention are intended to be included within the scope of protection of the present invention.

Claims

1. An intelligent design method for a multi-cellular structure with concave-convex minimal surface and adjustable dynamic mechanical response curve, characterized in that: The following steps are involved: Step S1: designing a wall concave-convex parameterized minimal surface unit cell structure based on a three-periodic minimal surface unit cell structure; Step S2: by changing the topological geometric parameters and concave-convex characteristic parameters of the unit cell structure of the minimal surface with concave-convex wall, a unit cell structure with different topological geometric parameters and concave-convex characteristics is formed, and a unit cell matrix is constructed; wherein the topological geometric parameters of the unit cell structure of the minimal surface with concave-convex wall are determined by the C value in the generalized governing equation of the three-periodic minimal surface: Cos(x)+Cos(y)+Cos(z)=C, and the value range of the C value is -1≤C≤1; the concave-convex characteristic parameters of the unit cell structure of the minimal surface with concave-convex wall include: concave-convex size, concave-convex depth and concave-convex distribution; Step S3: using an additive manufacturing method to prepare a test piece of a minimal surface unit cell structure with a concave-convex wall surface, and conducting experimental tests under experimental conditions of different loading rates; Step S4: performing a loading numerical simulation on the concave-convex minimal surface unit cell structure with topological geometric parameters and concave-convex features obtained in step S2, and obtaining a dynamic response result of the unit cell structure; Step S5: Based on the dynamic response results of the unit cell structure obtained in steps S3 and S4, a parameter matrix of topological geometric parameters and unit cell concavity and convexity, and a unit cell mechanical response matrix are constructed to form a training data set and a machine learning model is constructed for training, so that the machine learning model can predict the topological geometric parameters and unit cell concavity and convexity parameters of the unit cell structure of the minimal surface with concave and convex wall according to the mechanical response form; Step S6: combining the unit cell structures to form an N*N*N polycellular structure, performing experiments and simulations on the combined polycellular structure to obtain polycellular structure response data, performing machine learning classification and training using the unit cell structure combination as an input parameter set and the polycellular structure response data as a target set, to obtain a polycellular structure machine learning model with concave-convex minimal surfaces on the wall that can control the dynamic mechanical response curve; Step S7: Based on the prediction results of the machine learning model described in steps S5 and S6, the parameters of the single-cell structure and multi-cell structure of the concave-convex minimal surface of the wall that meet the actual engineering mechanics response target are obtained and the preparation is completed.

2. The intelligent design method for a multi-cellular structure with concave-convex minimal surface and adjustable dynamic mechanical response curve according to claim 1, characterized in that: The additive manufacturing method in step S3 includes any one of photocuring, melt extrusion, and selective laser melting.

3. The intelligent design method for a multi-cellular structure with concave-convex minimal surface and adjustable dynamic mechanical response curve according to claim 1, characterized in that: The loading rate of the experimental test in step S3 is 0.01 m / min~5 m / min.

4. The intelligent design method for a multi-cellular structure with concave-convex minimal surface and adjustable dynamic mechanical response curve according to claim 1, characterized in that: The dynamic response results in step S5 include: total absorbed energy, initial peak force, average crushing force, and stress-strain curve.

5. The intelligent design method for a multi-cellular structure with concave-convex minimal surface and adjustable dynamic mechanical response curve according to claim 1, characterized in that: The combination form of the N*N*N multicellular structure formed by combining the unit cell structures in step S6 is uniform, has an interlayer gradient, and is orthogonally anisotropic.

6. The intelligent design method for a multi-cellular structure with concave-convex minimal surface and adjustable dynamic mechanical response curve according to claim 1, characterized in that: The mechanical response targets required for the actual project in step S7 include one or more of energy absorption, peak force, and average crushing force targets.

Images