JP2023178008A - Design support system and design support method - Google Patents

Abstract

To provide a design support system capable of generating high-quality drawings from a density distribution of a structure derived from structural optimization calculations in a short time.SOLUTION: A design support system includes: a clustering condition input unit that receives input of a clustering condition; a clustering unit that follows the clustering condition; a machine learning control unit that performs machine learning on each cluster obtained; a drawing generation unit that generates new drawings from machine learning results; a drawing determination unit that determines whether a basic drawing to be compared and the new drawing are the same; an optimization/analysis condition input unit that receives input of calculation conditions for structural optimization; an analysis model generation unit that generates an analysis model according to calculation conditions; a structural optimization calculation unit that calculates using the analysis model; and a drawing generation condition input unit that displays the structural optimization calculation results and clustering results, and receives input of weights for each cluster required for drawing generation. The drawing generation unit generates a proposed drawing using the calculation results and weights of structural optimization.SELECTED DRAWING: Figure 1

Description

The present disclosure relates to a design support device and a design support method.

Conventionally, in design support technology for predicting the optimal shape of a mechanical structure, there is a method of predicting the optimal shape by using topology optimization calculation and image processing, which is a type of machine learning. Here, topology optimization is one of the structural optimization methods for deriving the optimal density distribution of a given structure using a given material based on the design variables of the structure. , also called phase optimization.

Patent Document 1 describes a modeling process that includes a learning method that enables the generation of physically realistic CAD data that can be directly used in the design and/or manufacturing process, the modeling process including a learning method that creates a dataset containing functional structures. (S10), and training of a generation autoencoder on a dataset using a learning method (S20), and may further include training of a latent space classifier using a learning method (S30), and one method using a generation method. or providing a plurality of latent vectors (S100) and generating a functional structure using a generation method (S200); It is disclosed that the method may include performing a conversion (S300).

Furthermore, in an apparatus for generating CAD drawings from the results of optimization calculations, there is a method of generating CAD drawings according to the model form of analysis. Here, CAD is an abbreviation for Computer Aided Design.

Patent Document 2 describes a finite element method model conversion device that converts CAD data into a finite element method model, an optimization device that constructs an optimized finite element method model using an optimization method, and a CAD converter that converts the optimized finite element method model. In a design optimization support system used for designing a structure, the system includes a CAD data conversion device that converts data into data, and a means for selecting a finite element method model creation method that is adapted to the type of CAD data and model form, and an optimization purpose. A system is disclosed that includes means for selecting an optimization method adapted to the above, and CAD data conversion means for converting an optimized finite element method model into CAD data according to the model form.

JP 2021-12693 Publication Japanese Patent Application Publication No. 2006-11729

In conventional topology design support technology, a tree representing the functional structure of a CAD drawing is machine learned using a deep neural network (DNN), and a new CAD drawing is generated from vector data called latent vectors. The optimal shape is predicted by performing topology optimization calculations using the created CAD drawings. In this case, the predicted optimal shape is represented by a density distribution that varies from 0 to 1 over the entire computational domain. That is, the optimal shape of the mechanical structure is predicted by assuming that there is no structure in the part where the density is 0 and that there is a structure in the part where the density is 1.

The optimal shape predicted in this way also includes density portions other than 0 and 1. In gray areas where the density is intermediate, such as 0.3 or 0.7, it is up to the engineer to determine whether a structure is included or not. For this reason, since the interpretation differs depending on the engineer, variations may occur in the CAD drawings that are finally created, and there is a problem that it is difficult to obtain CAD drawings with high quality content.

In the method described in Patent Document 1, sufficient consideration is not given to obtaining high-quality CAD drawings regardless of the engineer.

Furthermore, in the technique of generating a CAD drawing from the result of optimization calculation, the shape of the CAD drawing is determined by smoothly connecting the density distribution of the calculation area in the predicted optimal shape. Attributes of the CAD drawing include characteristic information such as manufacturing cost and efficiency.

In the design optimization support system described in Patent Document 2, the optimization objective is selected by the designer. Therefore, appropriately selecting an optimization objective from a plurality of characteristics etc. is left to the skill, experience, etc. of the designer, and it may be necessary to repeatedly review the optimization objective. Therefore, there is room for improvement in that the amount of effort and time required for design increases.

The present disclosure aims to generate high-quality drawings in a short time from the density distribution of a structure derived through structural optimization calculations.

The design support device of the present disclosure generates a proposal drawing from the calculation results of structure optimization that derives an optimized density distribution for a structure, and includes a clustering condition input section that receives input of clustering conditions, and a clustering condition input section that receives input of clustering conditions. a clustering unit that performs clustering based on the data, a machine learning control unit that performs machine learning on each cluster obtained by clustering, a drawing generation unit that generates a new drawing from the results of machine learning, and a basic drawing to be compared and a new drawing. a drawing discrimination section that compares drawings with other drawings and determines whether these drawings are the same; an optimization/analysis condition input section that accepts input of calculation conditions necessary for structural optimization calculations; There is an analytical model generation section that generates an analytical model, a structural optimization calculation section that performs structural optimization calculations using the analytical model, and a structural optimization calculation section that displays the structural optimization calculation results and clustering results and performs the calculations necessary for drawing generation. The drawing generation section includes a drawing generation condition input section that receives input of weights for each cluster, and a display section that displays information including at least one of the clustering conditions, calculation results, and generated drawings. A proposed drawing is generated using the structural optimization calculation results and weights, and the display unit displays the proposed drawing.

According to the present disclosure, it is possible to generate high-quality drawings in a short time from the density distribution of a structure derived through structural optimization calculations.

FIG. 1 is a schematic configuration diagram showing a design support device according to an embodiment. 7 is a flowchart showing clustering condition input processing in phase 1. FIG. 7 is a flowchart showing clustering processing in phase 1. 3 is a flowchart showing machine learning processing in phase 1. 7 is a flowchart showing topology optimization calculation processing in phase 2. 7 is a flowchart showing drawing generation processing in phase 2. It is a figure which shows an example of the data acquired in process S101 of FIG. 2A. 2A is a diagram showing an example of an input screen for clustering conditions used in step S102 of FIG. 2A. FIG. It is a figure which shows an example of the screen of the clustering result displayed in step S204 of FIG. 2B. It is a conceptual diagram showing an example of CycleGAN. FIG. 3 is a diagram showing an example of an input screen for topology optimization calculation. 4A is a diagram showing an example of the results of the topology optimization calculation obtained in step S403 of FIG. 4A. FIG. It is a figure which shows an example of the input screen of the generation conditions of a new drawing in step S502 of FIG. 4B. 4B is a diagram showing an example of a screen displayed in step S504 of FIG. 4B. FIG.

The present disclosure relates to a design support technology that generates a specific drawing based on input condition vectors (weights) from topology optimization results with density shading obtained through numerical analysis.

Examples will be described below with reference to the drawings.

FIG. 1 is a schematic configuration diagram showing a design support apparatus according to an embodiment.

As shown in this figure, the design support apparatus includes a clustering condition input section 101, a clustering section 102, a clustering result display section 103, a machine learning control section 104, a drawing generation section 105, a drawing discrimination section 106, Optimization/analysis condition input section 107, analysis model generation section 108, topology optimization calculation section 109 (structural optimization calculation section), drawing generation condition input section 110, drawing display section 111, database 112 (memory) device) and a computer 113.

In the clustering condition input section 101, an operator inputs information (clustering conditions) necessary for clustering, such as a CAD drawing, topology optimization calculation results, and characteristic information.

The clustering unit 102 performs clustering using a self-organizing map according to the input clustering conditions. Here, clustering is a type of machine learning, and refers to a method of dividing data into groups based on the degree of similarity between the data. Clustering is also referred to as "cluster analysis" or "data clustering."

The clustering result display section 103 displays the results of clustering using the self-organizing map, and also displays the average value of the characteristic information of each cluster.

The machine learning control unit 104 controls the drawing generation unit 105 and the drawing discrimination unit 106, and performs machine learning using the topology optimization calculation results, CAD drawings, and condition vectors as input using a method applying CycleGAN. Here, CycleGAN is an abbreviation for Cycle Generative Adversarial Networks.

The drawing generation unit 105 generates a new drawing using the CAD drawing and condition vector as input information as a result of the topology optimization calculation.

The drawing determination unit 106 uses the two drawings as input information to determine whether the two drawings are the same.

The optimization/analysis condition input unit 107 inputs analysis conditions, objective functions, and volume constraint conditions necessary for topology optimization calculations.

The analytical model generation unit 108 generates an analytical model based on the analytical conditions, objective function, and volume constraint conditions necessary for topology optimization calculation.

The topology optimization calculation unit 109 performs topology optimization calculations using the finite element method for the structural analysis part and the density method for the topology optimization part of the generated analytical model.

The drawing generation condition input unit 110 displays the results of the topology optimization calculation and the clustering results, and inputs the weight of each cluster necessary for drawing generation.

The drawing display section 111 displays the drawings generated by the drawing generation section 105 and the drawing generation conditions.

The database 112 stores data obtained from each section.

Next, an example of a processing procedure in the design support apparatus, that is, a design support method will be described.

The design support method of this embodiment can be broadly divided into two processes. The first process (phase 1) is a process of clustering data such as CAD drawings and subjecting the data to machine learning. The second process (phase 2) is a process of performing topology optimization calculations and generating drawings from the results.

Hereinafter, description will be made focusing on mechanical structures.

FIG. 2A is a flowchart showing clustering condition input processing in phase 1.

In steps S101 to S104 shown in the figure, conditions for clustering are input from the clustering condition input section 101 and stored in the database 112.

First, in step S101, past CAD drawings, topology optimization calculation results, characteristic information such as manufacturing cost, efficiency, strength performance, heat resistance performance, etc. are acquired from the database 112.

FIG. 5 is a diagram showing an example of data acquired in step S101.

As shown in this figure, the CAD drawing of "AA001" is related to the topology optimization calculation result "AA001TOP", and its attributes are linked to characteristic information such as manufacturing cost, efficiency, strength performance, and heat resistance. It is well managed. It is assumed that these past data are stored in the database 112 in advance. In this case, these past data are the source data for supervised learning. The past CAD drawings may be those obtained as a result of topology optimization calculations, but may also be drawings actually created by a designer in the past.

In step S102, the operator inputs conditions necessary for clustering using the clustering condition input section 101.

FIG. 6 is a diagram showing an example of an input screen for clustering conditions.

As shown in this figure, "mechanical structure 1" is input as the learning model name. As the variables to be clustered, the data attributes shown in FIG. 5 are displayed in a corresponding manner. In Figure 6, CAD drawings, topology optimization calculation results, manufacturing cost, efficiency, strength resistance performance, and heat resistance performance are displayed, and manufacturing cost, efficiency, strength resistance performance, and heat resistance performance are selected as clustering conditions. There is. It is also possible to input the number of clusters (number of clusters) for dividing the acquired data into clusters. Here, "10" is input as the number of clusters.

In step S103, information such as the learning model name, variables serving as clustering conditions, and number of clusters input in step S102 is acquired.

In step S104, the information acquired in step S103 is registered in the database 112.

FIG. 2B is a flowchart showing clustering processing in phase 1.

In steps S201 to S205 shown in the figure, the clustering unit 102 divides the data into clusters, and the clustering result display unit 103 displays the clustered results.

In step S201, the clustering unit 102 acquires past CAD drawings, topology optimization calculation results, characteristic information such as manufacturing cost, efficiency, strength resistance performance, heat resistance performance, etc. from the database 112.

In step S202, the clustering unit 102 obtains the information obtained in steps S101 to S104. Here, the learning model name, the variable to be clustered, and the number of clusters are acquired.

In step S203, the clustering unit 102 performs clustering. Various clustering methods have been proposed. Here, clustering is performed using a method called self-organizing map. A self-organizing map is a type of neural network that models the visual cortex of the cerebral cortex. In the self-organizing map, weight vectors are randomly arranged on the map and one input vector is prepared. The similarity with the input vector is calculated for all weight vectors on the map. Euclidean distance is used for similarity. The vector with the smallest distance between each vector is found, and the weight vectors in its vicinity are changed using the following equation (1).

In the formula, W _u is a weight vector, θ is a neighborhood radius, and α is a learning coefficient. U is an input vector containing variables to be clustered. Further, n represents the number of repetitions. In this way, using a self-organizing map, it is divided into clusters with high similarity. Here, it is divided into 10 clusters.

In step S204, the clustering result display section 103 displays the clustered results.

FIG. 7 is a diagram showing an example of a display screen of the clustering results.

In this figure, variables to be clustered such as manufacturing cost, efficiency, strength resistance performance, heat resistance performance, etc. are divided and displayed as highly similar clusters from "1" to "10". Further, the table shown in this figure displays information such as the average value of variables such as manufacturing cost, efficiency, strength resistance performance, heat resistance performance, etc. of each cluster. This allows the characteristics of the similarity of each cluster to be grasped, such as cluster "1" having low manufacturing cost and cluster "3" having high efficiency. If the operator is satisfied with the result, the operator presses the "determination" button, and if the operator wants to try again, the operator presses the "redo" button to review the clustering conditions from step S101.

In step S205, the results of clustering in step S203 are registered in the database 112. Here, clustering information divided into 10 variables and information on the average value of each variable are registered.

FIG. 3 is a flowchart showing machine learning processing in phase 1.

The machine learning shown in this figure is executed by controlling the drawing generation section 105 and the drawing discrimination section 106 by the machine learning control section 104. Various machine learning methods have been proposed, but here, machine learning is performed by applying an image processing method called CycleGAN.

FIG. 8 is a conceptual diagram showing an example of CycleGAN.

CycleGAN generates new images from images of horses, etc. For example, if a zebra is used as a real image, it is determined whether the newly generated image is the same as the zebra image, and machine learning is performed to adjust the image generation parameters so that they are the same. At this time, a new image is generated from the real image of the zebra, and the image is determined to see if the newly generated image is the same as the horse image. Machine learning is performed to ensure that they are the same. Adjust parameters. These machine learning processes are repeated until they are determined to be the same.

In this way, CycleGAN is characterized in that it can generate an image resembling a zebra from an image of a horse or the like. At this time, by using a condition vector, it is possible to generate an image resembling a zebra and an image resembling a deer from an image of a horse. That is, for example, if the real image is a zebra, input the condition vector [1 0 0 0 0] for machine learning, and if the real image is a deer, input the condition vector [0 1 0 0 0]. machine learning. If you want to generate an image resembling a zebra after machine learning, input the image of a horse and the condition vector [1 0 0 0 0] to generate the image, and an image resembling a zebra will be generated. Furthermore, by inputting the condition vector [1 1 0 0 0], an image resembling a zebra and a deer can also be generated.

A neural network is used to generate images. A neural network is a mathematical model that aims to express the characteristics of the brain, which is composed of a large number of neurons, through computer simulation. The neural network is given by a recurrence formula such as the following formula (2), where each layer of artificial neurons is represented by X _i .

In the formula, A _i is a weight parameter and B _i is a bias parameter. f is an activation function. Determine A _i and B _i through machine learning. Note that in the case of three layers, _X1 is the input layer, _X2 is the intermediate layer, and _X3 is the output layer. A network with multiple intermediate layers is called a deep neural network. Deep neural networks are applied to CycleGAN image processing.

In step S301, the machine learning control unit 104 retrieves past CAD drawings, topology optimization calculation results, clustering conditions obtained in the step shown in FIG. 2A, and all clustering data obtained in the step shown in FIG. 2B from the database 112 by the machine learning control unit 104. get.

In step S302, one of the clusters clustered by the machine learning control unit 104 is extracted, and the result of the topology optimization calculation and the condition vector belonging to this cluster are input to the drawing generation unit 105 to generate a new drawing. Here, since there are 10 clusters, a new drawing is generated by inputting the condition vector [1 0 0 0 0 0 0 0 0 0] as a result of topology optimization calculation belonging to cluster number "1".

In step S303, the new drawing generated in step S302 by the machine learning control unit 104 and a drawing related to the input result of the topology optimization calculation are input to the drawing discrimination unit 106, and the new drawing and Determine whether the drawings related to the input topology optimization calculation results are the same. If it is determined that they are not the same, the parameters of the drawing generation unit 105 are adjusted. Here, the drawings related to the results of the input topology optimization calculations will be compared as "real drawings" and will also be referred to as "basic drawings."

In step S304, the machine learning control unit 104 inputs the real drawing and condition vector used in step S303 to the drawing generation unit 105 to generate a new drawing. The condition vector [1 0 0 0 0 0 0 0 0 0] is used here as well.

In step S305, the new drawing generated in step S304 by the machine learning control unit 104 and the result of the topology optimization calculation used in step S302 are input to the drawing discrimination unit 106, and the new drawing and the calculation result are input to the drawing determination unit 106. are the same. If it is determined that they are not the same, the parameters of the drawing generation unit 105 are adjusted. Here, whether or not they are the same is determined by machine learning. Specifically, the determination is made using a neural network.

In step S306, the machine learning control unit 104 performs machine learning in steps S302 to S305 on all topology optimization calculation results and drawings of data belonging to the cluster, and repeats the process until convergence. Furthermore, machine learning is similarly performed on all of the data in each of the remaining clusters and is repeated until convergence. Taking the case of performing machine learning on the data of cluster number "3" as an example, machine learning is performed using the condition vector [0 0 1 0 0 0 0 0 0 0]. Here, the convergence determination condition is usually an arbitrary condition that is accepted by the user. For example, if the user input value is "99%", it is assumed that the images have converged and the calculation is terminated when the images are determined to be "the same" with a probability of 99% or more. If the input value of ``99%'' is too high and convergence cannot be reached, the user can forcibly terminate the calculation, relax the convergence criteria, and re-enter ``95%'' for calculation. . Roughly 95% of the input values are considered appropriate.

In step S307, information on the results of machine learning performed by the machine learning control unit 104 is registered in the database 112. Here, machine learning information with parameters adjusted so that the generated drawing is the same as the original drawing is registered in the database.

The above is Phase 1.

Next, phase 2 will be explained.

FIG. 4A is a flowchart showing topology optimization calculation processing in phase 2.

In the topology optimization calculation process shown in this figure, the conditions necessary for the topology optimization calculation are input by the optimization/analysis condition input unit 107, and the conditions necessary for the topology optimization calculation are input by the optimization/analysis condition input unit 107 by the analysis model generation unit 108. An analytical model is generated according to the conditions, and the topology optimization calculation unit 109 performs topology optimization calculation for the mechanical structure.

In step S401, the optimization/analysis condition input unit 107 receives optimization conditions and analysis conditions input by the operator.

FIG. 9 is a diagram showing an example of an input screen.

In this figure, "hook" is input as the analysis model name. An analysis model for topology optimization calculation is created by the operator. In the calculation area, which is a rectangular area, a boundary condition in which the left side in the figure is constrained by a wall is given, and a load condition of 1000 N is given in the center on the right side in the figure. Furthermore, "aluminum" is input as the material condition. As the conditions for the topology optimization calculation, a condition that minimizes the average compliance represented by the product of load and displacement is input, and "40% or less" is input as the volume constraint condition. That is, optimization conditions are set such that the volume of the rectangular area is 40% or less and the rigidity is maximized at that time.

In step S402, the analytical model generation unit 108 generates an analytical model based on the conditions input in step S401. The computational domain is divided into meshes called finite elements, and an analytical model containing information on the analytical model, analysis and optimization conditions, etc. is generated.

In step S403, the topology optimization calculation unit 109 inputs the analytical model generated in step S402 and performs topology optimization calculation. Here, the finite element method is used for the structural analysis part, and the density method is used for the topology optimization part. In this calculation method, each finite element is provided with a density varying from 0 to 1 as a design variable, and optimization is performed by calculating the product of the density and Young's modulus as the Young's modulus of the finite element. Therefore, a portion with a density of 0 means a state in which no structure is contained, and a portion with a density of 1 means a state in which a structure is contained.

FIG. 10 is a diagram showing an example of the results of topology optimization calculation. In the figure, the shading of the divided calculation areas represents the level of density. The darker the area, the higher the density.

As this figure shows, the structure of a mechanical structure is expressed by the shading of the calculation domain.

FIG. 4B is a flowchart showing drawing generation processing in phase 2.

In the drawing generation process shown in this figure, the drawing generation condition input unit 110 inputs conditions for generating a new drawing from the results of topology optimization calculation, and the machine learning control unit 104 inputs the conditions to the drawing generation condition input unit 110. The generated information is input to the drawing generation section 105 to generate a new drawing, and the drawing display section 111 displays the generated drawing.

In step S501, the machine learning control unit 104 acquires all information input in the steps shown in FIGS. 2A, 2B, and 3.

In step S502, the operator inputs conditions for generating a new drawing from the results of the topology optimization calculation using the drawing generation condition input unit 110.

FIG. 11 is a diagram showing an example of an input screen for generating conditions for a new drawing.

The operator generates a new drawing by inputting an analysis model name and a learning model name, and inputting drawing generation conditions. In this figure, "hook" is input as the analysis model name, and "mechanical structure 1" is displayed as the learning model name.

In the topology optimization calculation result display area, the topology optimization calculation results shown in FIG. 4A are displayed.

Furthermore, the clustering results shown in FIG. 2B are displayed in the lower display area of FIG. 11. The operator generates a new drawing by changing the weight of each cluster number between 0 and 1. Cluster number "3" has a weight of "1", cluster number "5" has a weight of "0.5", and cluster number "7" has a weight of "1". The weights of other clusters are set to "0".

In step S503, the machine learning control unit 104 inputs the information input in step S502 to the drawing generation unit 105 to generate a new drawing. Here, as a result of the topology optimization calculation for the analysis model name "Hook", "1", "0.5", and "1" are included in the elements corresponding to cluster numbers "3", "5", and "7". The condition vector [0 0 1 0 0.5 0 1 0 0 0] is input to the drawing generation unit 105 to generate a new drawing. In this way, a new screen is generated in which the properties of cluster numbers "3", "5", and "7" have the respective properties in a ratio of 1:0.5:1.

In step S504, the drawing display unit 111 displays the drawing generated in step S503.

FIG. 12 is a diagram showing an example of the generated drawing display screen.

In this figure, the analysis model name "Hook" and the learning model name "Machine Structure 1" are displayed. Further, the drawing generated in step S503 (proposed drawing) and the drawing generation conditions input in step S502 are displayed.

In step S505, the drawing generated as a result of the topology optimization calculation shown in FIG. 4A is registered in the database 112. Note that after displaying the drawing in step S504, the process may return to step S502 and input by the operator may be repeated, if necessary.

In this way, past CAD drawings and topology optimization calculation results are divided into clusters with high similarity based on characteristic information such as manufacturing cost and efficiency, and machine learning is applied to each cluster to improve topology optimization calculations. Generate a new drawing from the results. As a result, a shape that does not depend on the engineer's experience can be obtained, compared to the shape that the engineer had conventionally predicted based on the density density obtained from experience in topology optimization calculations. Furthermore, new drawings can be generated that take into consideration characteristics other than the objective function of optimization calculations, such as manufacturing cost and efficiency. Furthermore, by applying condition vectors, drawings that take into account multiple characteristics such as manufacturing cost and efficiency can be generated in a short time.

As described above, according to the present disclosure, high-quality drawings can be obtained in a short time.

In this disclosure, clustering, machine learning, and topology optimization calculations are described as being performed on the same computer (computer), but they can be performed on different computers by using a network environment. It is. That is, the design support method according to the present disclosure may be implemented by configuring a cloud by having a plurality of servers installed in remote locations cooperate using the Internet. In that case, it can be said that a plurality of servers etc. constitute a design support system.

101: Clustering condition input section, 102: Clustering section, 103: Clustering result display section, 104: Machine learning control section, 105: Drawing generation section, 106: Drawing discrimination section, 107: Optimization/analysis condition input section, 108: Analysis model generation section, 109: Topology optimization calculation section, 110: Drawing generation condition input section, 111: Drawing display section, 112: Database, 113: Computer.

Claims (14)

A design support device that generates a proposal drawing from structural optimization calculation results that derive an optimized density distribution for a structure,
a clustering condition input section that accepts input of clustering conditions;
a clustering unit that performs clustering according to the clustering condition;
a machine learning control unit that performs machine learning on each of the clusters obtained by the clustering;
a drawing generation unit that generates a new drawing from the result of the machine learning;
a drawing determination unit that compares the basic drawing to be compared with the new drawing and determines whether these drawings are the same;
an optimization/analysis condition input unit that receives input of calculation conditions necessary for the structural optimization calculation;
an analytical model generation unit that generates an analytical model according to the calculation conditions;
a structural optimization calculation unit that performs the calculation of the structural optimization using the analytical model;
a drawing generation condition input unit that displays the calculation results of the structural optimization and the clustering results and receives input of weights for each of the clusters necessary for drawing generation;
a display unit that displays information including at least one of the clustering conditions, the calculation results, and the generated drawings;
The drawing generation unit generates a proposed drawing using the calculation result of the structural optimization and the weight,
The display unit is a design support device that displays the proposed drawing. The design support apparatus according to claim 1, wherein the clustering condition includes at least one of manufacturing cost and efficiency of the structure. The design support apparatus according to claim 1, wherein the display section displays information including characteristic information of each of the clusters. 2. The design support apparatus according to claim 1, wherein the calculation of the structural optimization uses a finite element method and a density method. 2. The design support apparatus according to claim 1, wherein the clustering is performed using a self-organizing map. The design support apparatus according to claim 1, wherein the structural optimization is topology optimization. 7. The design support apparatus according to claim 6, wherein the drawing generation unit generates the proposed drawing by the machine learning. A design support method for generating a proposal drawing from structural optimization calculation results for deriving an optimized density distribution for a structure, the method comprising:
The clustering condition input section accepts input of clustering conditions,
a clustering unit performs clustering according to the clustering conditions,
a machine learning control unit performs machine learning on each of the clusters obtained by the clustering,
a drawing generation unit generates a new drawing from the result of the machine learning,
a drawing discrimination unit compares the basic drawing to be compared with the new drawing and determines whether these drawings are the same;
The optimization/analysis condition input unit receives input of calculation conditions necessary for the structural optimization calculation,
an analytical model generation unit generates an analytical model according to the calculation conditions,
a structural optimization calculation unit performs the calculation of the structural optimization using the analytical model,
a drawing generation condition input unit displays the calculation results of the structural optimization and the clustering results, and receives input of weights for each of the clusters necessary for drawing generation;
a display unit displays information including at least one of the clustering condition, the calculation result, and the generated drawing;
The drawing generation unit generates a proposed drawing using the calculation result of the structural optimization and the weight,
The design support method, wherein the display unit displays the proposed drawing. 9. The design support method according to claim 8, wherein the clustering condition includes at least one of manufacturing cost and efficiency of the structure. 9. The design support method according to claim 8, wherein the display unit displays information including characteristic information of each of the clusters. 9. The design support method according to claim 8, wherein the calculation of the structural optimization uses a finite element method and a density method. 9. The design support method according to claim 8, wherein the clustering is performed using a self-organizing map. 9. The design support method according to claim 8, wherein the structural optimization is topology optimization. 14. The design support method according to claim 13, wherein the drawing generation unit generates the proposed drawing by the machine learning.

Images