JP2008046736A - System and method for generating model for analysis - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily generate a mesh model, which is a model for analysis, even in case of a form model having a complicated inner structure such as a liquid crystal television. <P>SOLUTION: Form models, which are objects to be analyzed by a two-dimensional face model or a three-dimensional shell model which both are generated by a form model input device 10, are stored in a form model database 20. The form models stored in the form model database 20 are rectangularly divided by a form model processing part 30. Each of the forms of the division models, which are rectangularly divided, is simplified by using basic parts, and the simplified division model is mesh-divided and combined with other models by a mesh division processing part 40 to obtain a mesh model which is a model for analysis. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  In the present invention, a shape model as an analysis object is divided into rectangles to form a plurality of blocks, and then the shape of each block division model is recognized by an information processing model such as a neural network, and the division model of each block is determined. By selecting the basic part that is the closest to the shape and simplified in advance, the shape of each block division model is simplified, and each simplified basic part is divided into meshes. The present invention relates to an analysis model generation system and an analysis model generation method for obtaining a mesh model as an analysis model by combining them.

  As a method of creating a mesh model, which is an analysis model, from a shape model (two-dimensional surface model or three-dimensional shell model), normalize the shape model, which is the analysis target, into a set of rectangular areas using a mathematical method. A method is often used in which meshing suitable for an analysis program is applied to the shape model and the meshing is mapped to the original shape model.

  In addition, the operator interactively adds an auxiliary line for instructing an appropriate block division to the shape model, divides the shape model into blocks of a plurality of triangles or quadrilateral regions, and concretely specifies the ridge lines of each region. There is also a method of creating a mesh model by giving the number of mesh divisions.

  However, in the method of creating a mesh model by the method based on the former mapping, there is a limit to the boundary shape that can be reproduced. That is, in the case of a shape model having special graphic features such as cracks or convex curves, it is difficult to accurately recognize and normalize those graphic features, and the normalized mesh is mapped to the original shape model. Sometimes, the mesh shape is locally distorted. This distortion becomes a factor of lowering the analysis accuracy.

  In addition, the method in which the latter operator interactively adds auxiliary lines to the shape model and divides the shape model into a plurality of simple regions reflects the operator's empirical elements, and thus has the above-mentioned special graphic features. Although a mesh model with relatively high analysis accuracy can be created for the shape, it is necessary to divide the shape in consideration of the aspect ratio of the mesh, and the burden on the operator is very large.

  Therefore, Patent Document 1 discloses a modeling system that can subdivide a shape model having special graphic features without causing local distortion in the mesh shape, and can greatly reduce the burden on the operator. is suggesting.

This is because the shape recognizing means uses geometric features such as the position, line length, line type, and inter-line angle of the line segment constituting the shape model, the number of line segments constituting the shape model, the number of holes or holes, the line Recognize overall features such as topological features such as the coupling relationship of the minute and the inclination of the shape model, the presence or absence of cracks, the presence or absence of interphase lines, and shape division for the recognized shape model by the shape model division processing means A line segment is added to the shape model by applying the division rule selected from the rule database, and the shape model is automatically divided into a plurality of quadrilaterals or triangle parts.
Japanese Patent No. 3138933

  As described above, in Patent Document 1, even for a shape model having special graphic features, by applying a division rule selected from the shape division rule database and adding a line segment to the shape model, the burden on the operator is reduced. It is possible to generate a mesh model with high analysis accuracy while greatly reducing.

  However, the shape division rule of the shape division rule database in Patent Document 1 includes a local division rule for dividing a partial shape such as an arc, a free curve, a triangle, a pentagon, etc. that constitutes a shape, and an interpenetrating point or division process. Consists of a basic division rule to eliminate the point on the line that occurs, a special division rule to effectively divide and capture the characteristics of the entire shape, and a division restriction rule to suppress the addition of unauthorized auxiliary lines Has been.

  Therefore, it is effective when the shape model to be analyzed is a relatively simple shape, but in the case of a shape model having a complicated internal structure such as a liquid crystal television, these shape division rules are applied. When trying to divide the shape, the number of shape divisions increases too much, which makes it extremely difficult to generate a mesh model.

  The problem to be solved is that it is possible to generate a mesh model with high analysis accuracy by using the shape division rule. For example, in the case of a shape model having a complicated internal structure such as a liquid crystal television, The number of shape divisions increases too much, and the generation of the mesh model becomes extremely difficult.

The analysis model generation system of the present invention simplifies the shape of the basic shape model and the shape model database that stores the shape model that is the analysis object of the two-dimensional surface model or the three-dimensional shell model created by the shape model input device. The basic part database that stores the converted basic parts, the shape model that is the analysis object is divided into rectangles, and the basic parts of the basic shape models that correspond to the shape recognition of each divided divided model are used as the basic model database. Shape model processing means for simplifying the division model and mesh division processing means for obtaining a mesh model as an analysis model by combining the simplified individual division models by mesh division.
The shape model processing means performs shape recognition of a shape model rectangle dividing means that divides the shape model, which is the analysis object, into a plurality of blocks, and a divided model of each of the divided blocks. The shape of the divided model of each block is selected by selecting from the basic part database the divided shape recognizing means and the basic part whose shape is closest to the shape of the divided model of each block whose shape has been recognized. Basic part selection means that simplifies the process can be provided.
In addition, every time shape recognition of the divided model is performed, a shape corresponding to the shape recognition can be stored in the basic part database as the basic part.
The analysis model generation method of the present invention includes a shape model storing step for storing a shape model that is an analysis object of a two-dimensional surface model or a three-dimensional shell model created by a shape model input device, and a shape of a basic shape model. A basic part storing step for storing simplified basic parts, and a shape to be simplified by dividing the shape model into rectangles, and using the basic parts of the basic shape model according to shape recognition of each divided divided rectangle model The method includes a model processing step, and a mesh division processing step of obtaining a mesh model as an analysis model by combining the simplified individual division models by dividing the mesh.
Further, a shape model rectangle dividing step of dividing the shape model into a plurality of blocks, a divided shape recognition step for recognizing the shape of the divided model of each of the divided blocks, and each of the shape recognized A basic part selection step for simplifying the shape of the divided model of each block by selecting the basic part that is closest to the shape of the divided model of the block and selecting the basic part having the simplified shape from the basic part database. Can be.
In addition, every time shape recognition of the divided model is performed, a shape corresponding to the shape recognition can be stored as the basic part.
In the analysis model generation system and the analysis model generation method of the present invention, the shape model database stores a shape model that is an analysis object of the two-dimensional surface model or the three-dimensional shell model created by the shape model input device, The shape model processing means divides the shape model stored in the shape model database into rectangles, simplifies the shape of each divided rectangle model using basic parts, and the mesh division processing means simplifies individual shapes. A mesh model, which is an analysis model, is obtained by dividing and combining the divided models.

  According to the analysis model generation system and the analysis model generation method of the present invention, a shape model that is an analysis object of a two-dimensional surface model or a three-dimensional shell model created by a shape model input device is stored in the shape model database. Then, the shape model stored in the shape model database is divided into rectangles by the shape model processing means, and the shape of each divided divided model is simplified using basic parts, and simplified by the mesh division processing means. A mesh model that is an analysis model can be obtained by combining individual divided models into meshes. For example, even in the case of a shape model having a complicated internal structure such as a liquid crystal television, a rectangular model is used. Dividing reduces the number of shape divisions to the minimum, and the shape of each division model is the basic pattern. From being simplified using a tool, it is possible to easily generate a mesh model.

  In the present embodiment, the shape model database stores the shape model that is the analysis target of the two-dimensional surface model or the three-dimensional shell model created by the shape model input device, and the shape model processing means stage stores the shape model in the shape model database. By dividing the stored shape model into rectangles, simplifying the shape of each divided rectangle model using basic parts, and combining the individual divided models with mesh division by mesh division processing means A mesh model that is an analysis model can be obtained. Even in the case of a shape model having a complicated internal structure such as a liquid crystal television, the number of shape divisions can be minimized by rectangular division. The mesh model by simplifying the shape of the split model using basic parts And so it can be easily performed.

Embodiments of the present invention will be described below.
FIG. 1 is a block diagram showing an embodiment of an analysis model generation system of the present invention, FIG. 2 is a flowchart for explaining an analysis model generation method by the analysis model generation system of FIG. 1, and FIG. FIG. 4 is a diagram for explaining learning for selecting a basic part in which the shape of the basic shape model is simplified by the analysis model generation system of FIG. 1, and FIG. 4 is an analysis model generation by the analysis model generation system of FIG. It is a figure for demonstrating a method.

  The analysis model generation system shown in FIG. 1 divides a shape model 22 to be analyzed, which will be described later, into a plurality of blocks by dividing the shape into a plurality of blocks, and then a later-described divided model 22a for each block by an information processing model such as a neural network. By recognizing ~ 22g shapes, and selecting the following basic parts 23a-23g, which are the closest to the shapes of the following divided models 22a-22g of the respective blocks and whose shapes obtained by learning in advance are simplified. The shape of the divided models 22a to 22g to be described later of each block is simplified, and the mesh model 24 to be described later, which is an analysis model, is obtained by dividing and combining the simplified individual basic parts 23a to 23g to be described later. Is.

  The analysis model generation system includes a shape model input device 10, a shape model database 20, a basic part database 21, a shape model processing unit 30, a mesh division processing unit 40, a mesh model database 50, and a mesh model display processing unit 60. .

    The shape model input device 10 includes a display, a keyboard, a mouse, and the like, receives input of data of a shape model 22 described later, which is an analysis object, and creates a two-dimensional surface model or a three-dimensional shell model.

  The shape model database 20 stores a later-described shape model 22 which is an analysis object of a two-dimensional surface model or a three-dimensional shell model created by the shape model input device 10.

  The basic parts database 21 has basic parts (a) -2 to (to be described later) in which the shapes of basic shape models (a) -1 to (c) -1 to be described later are simplified by learning using an information processing model such as a neural network. c) Store -2. Details of the basic parts (a) -2 to (c) -2 will be described later.

  The shape model processing unit 30 as a shape model processing means is an information processing model such as a neural network. After the analysis object is divided into a plurality of blocks by dividing the object into rectangles, divided models 22a to 22g, which will be described later, of the respective blocks. Each block is recognized by performing shape recognition and selecting a later-described basic part 23a to 23g whose shape obtained by learning in advance is closest to the shape of a later-described divided model 22a to 22g of each block. The shape of a division model 22a to 22g, which will be described later, is simplified, and includes a shape model rectangular division unit 31, a division shape recognition unit 32, and a basic part selection unit 33.

  The shape model rectangle dividing unit 31 as the shape model rectangle dividing unit divides a shape model 22 described later, which is an analysis object stored in the shape model database 20, into a rectangular shape, details of which will be described later. To do.

  The divided shape recognizing unit 32 as a divided shape recognizing unit recognizes the shapes of divided models 22a to 22g (to be described later) of each block divided by the shape model rectangular dividing unit 31, and details thereof will be described later. .

  The basic part selecting unit 33 as the basic part selecting unit stores, from the basic parts database 21, the below-described basic parts 23a to 23g that are close to the shapes of the below-described divided models 22a to 22g of each block recognized by the divided shape recognition unit 32. By selecting and replacing the selected later-described basic parts 23a to 23g with later-described divided models 22a to 22g of the divided blocks, the shapes of the later-described divided models 22a to 22g are simplified. Details will be described later.

  The mesh division processing unit 40 as the mesh division processing means divides the individual later-described basic parts 23a to 23g selected by the basic part selection unit 33 into meshes, and combines the later-described mesh model 24, which is an analysis model. obtain. In addition, what is necessary is just to set the number in the case of a mesh division | segmentation arbitrarily according to the shape of the below-mentioned mesh model 24 which is a model for analysis.

  The mesh model database 50 stores a mesh model 24 to be described later, which is an analysis model obtained by the mesh division processing unit 40.

  The mesh model display processing unit 60 performs display processing of a mesh model 24 described later, which is an analysis model stored in the mesh model database 50. Note that a mesh model 24 described later that has been subjected to display processing by the mesh model display processing unit 60 may be displayed on the display of the shape model input device 10 or may be displayed on another display. .

  Here, the mesh model 24 described later is stored in the mesh model database 50 and then the display processing of the mesh model 24 described later by the mesh model display processing unit 60 is performed. It goes without saying that a mesh model 24 (to be described later) may be displayed on the display so that an operation such as correction can be performed.

  Next, the flow of the processing procedure of the analysis model generation method will be described. In addition, although the basic parts (a) -2 to (c) -2 and the basic parts 23a to 23g having different reference numerals are different in shape from each other, they are the same, and are used properly according to the flow of explanation. It shall be.

  First, as shown in FIGS. 2 and 3, in the case of the basic shape model (a) -1 with respect to an information processing model such as a neural network, basic parts in which the shape of the basic shape model (a) -1 is simplified. (A) Learn to select -2 (step S1).

  In the case of the basic shape model (b) -1 with respect to the information processing model such as a neural network, the basic part (b) -2 in which the shape of the basic shape model (b) -1 is simplified is selected. Let them learn.

  Further, in the case of the basic shape model (c) -1 for an information processing model such as a neural network, the basic part (c) -2 in which the shape of the basic shape model (c) -1 is simplified is selected. Let them learn.

  The basic parts (a) -2 to (c) -2 learned by the information processing model such as a neural network are stored in the basic parts database 21 (step S2).

  Next, as shown in FIG. 4, the shape model rectangle dividing unit 31 divides the shape model 22 as the analysis object into a rectangular shape (step S3). Here, the shape model 22 is, for example, a two-dimensional surface model created by the shape model input device 10, and is stored in the shape model database 20.

  FIG. 4 shows a case where the shape model 22 is divided into seven blocks. In addition, what is necessary is just to set arbitrarily the division | segmentation number in this case according to the shape of the shape model 22 which is an analysis object.

  Next, the divided shape recognition unit 32 recognizes the shapes of the divided models 22a to 22g of the divided blocks (step S4). Here, when the shapes of the division models 22a to 22g of each block are recognized, the basic parts (a) -2 to (c) close to the shapes of the division models 22a to 22g of each block are detected by the basic part selector 33. ) -2 is selected from the basic part database 21 (step S5), and the divided model 22a of each block into which any of the selected basic parts (a) -2 to (c) -2 is divided. It is replaced with ~ 22g (step S6).

  That is, the divided model 22a is replaced with the basic part 23a, the divided model 22b is replaced with the basic part 23b, the divided model 22c is replaced with the basic part 23c, the divided model 22d is replaced with the basic part 23d, and the divided model 22e is The divided model 22f is replaced with the basic part 23e, the divided model 22f is replaced with the basic part 23f, and the divided model 22g is replaced with the basic part 23g.

  By replacing the divided models 22a to 22g of each block with the basic parts 23a to 23g, the shapes of the divided models 22a to 22g of each block are simplified (step S7).

  Next, the mesh division processing unit 40 meshes and combines the individual basic parts 23a to 23g selected by the basic part selection unit 33 (step S8), thereby obtaining a mesh model 24 as an analysis model. (Step S9). In addition, what is necessary is just to set the number in the case of a mesh division | segmentation arbitrarily according to the shape of the mesh model 24 which is an analysis model.

  The mesh model 24, which is an analysis model obtained by the mesh division processing unit 40, is stored in the mesh model database 50 (step S10).

  Next, the mesh model display processing unit 60 performs display processing of the mesh model 24, which is an analysis model stored in the mesh model database 50 (step S11), and displays it on the display of the geometric model input device 10 or other display. It is displayed (step S12).

  As described above, in this embodiment, the shape model database 20 stores the shape model 22 that is the analysis target of the two-dimensional surface model or the three-dimensional shell model created by the shape model input device 10, and the shape model processing unit 30, the shape model 22 stored in the shape model database 20 is divided into rectangles, and the shapes of the divided models 22 a to 22 g divided into rectangles are simplified using the basic parts 23 a to 23 g. The mesh model 24, which is an analysis model, can be obtained by dividing and combining the individual divided models 22a to 22g.

  Specifically, the shape model processing unit 30 performs shape recognition of the shape model rectangle dividing unit 31 that divides the shape model 22 into a plurality of blocks, and the divided models 22a to 22g of each of the divided blocks. By selecting, from the basic parts database 21, the divided shape recognition unit 32 to perform and the basic parts 23a to 23g that are closest to the shapes of the divided models 22a to 22g of the respective blocks whose shapes are recognized and whose shapes are simplified, respectively. And a basic part selection unit 33 that simplifies the shape of the block division models 22a to 22g, and the mesh division processing unit 40 is configured so that the simplified individual basic parts 23a to 23g are mesh-divided and combined. did.

  As a result, even in the case of the shape model 22 having a complicated internal structure or the like such as a liquid crystal television, the number of shape divisions is minimized by the rectangular division, and the shape of each of the division models 22a to 22g is reduced. Since the basic parts 23a to 23g are used for simplification, the mesh model 24 can be easily generated.

  In the present embodiment, the shape recognition of the divided models 22a to 22g of each block divided into rectangles by the shape model rectangle dividing unit 31 is performed by the divided shape recognizing unit 32, and the shape is recognized by the basic part selecting unit 33. Although the case where the basic parts 23a to 23g that are closest to the shape of the division models 22a to 22g of each block and whose shapes are simplified and obtained in advance by learning is selected from the basic parts database 21 has been described. The basic parts 23 a to 23 g with simplified shapes are stored in the basic part database 21 every time the divided models 22 a to 22 g are recognized by the divided shape recognizing unit 32, and the basic parts selecting unit 33 is used. May select the stored basic parts 23a to 23g.

  In this case, every time shape recognition of the division models 22a to 22g is performed by the division shape recognition unit 32, the basic parts 23a to 23g are stored in the basic part database 21, so that the accumulation amount of the basic parts 23a to 23g increases. It can be used for the subsequent rectangular division.

  The present invention can be applied to a method for obtaining a mesh model, which is an analysis model for any analysis target, such as structural analysis or fluid analysis.

It is a block diagram which shows one Embodiment of the model generation system for analysis of this invention. 3 is a flowchart for explaining an analysis model generation method by the analysis model generation system of FIG. 1. It is a figure for demonstrating the learning for selecting the basic part which simplified the shape of the basic shape model by the model generation system for analysis of FIG. It is a figure for demonstrating the analysis model production | generation method by the analysis model production | generation system of FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Shape model input device 20 Shape model database 21 Basic part database 22 Shape model 22a-22g Division | segmentation model 23a-23g Basic part (a) -2-(c) -2 Basic part 30 Shape model process part 31 Shape model rectangle division | segmentation part 32 division shape recognition unit 33 basic part selection unit 40 mesh division processing unit 50 mesh model database 60 mesh model display processing unit

Claims (6)

A shape model database for storing a shape model which is an analysis object of a 2D surface model or a 3D shell model created by the shape model input device;
A basic parts database that stores basic parts with simplified basic shape models,
A shape model processing means that divides the shape model as the analysis object into rectangles, and simplifies the divided model using basic parts of the basic shape model according to shape recognition of each divided model divided into rectangles;
An analysis model generation system comprising: a mesh division processing unit that obtains a mesh model that is an analysis model by dividing and combining the simplified individual division models. The shape model processing means includes:
A shape model rectangle dividing means that divides the shape model that is the analysis object into a plurality of blocks by dividing the shape model;
Divided shape recognition means for recognizing the shape of a divided model of each of the rectangular divided blocks;
A basic part that simplifies the shape of the divided model of each block by selecting the basic part that is closest to the shape of the divided model of each block whose shape has been recognized and whose shape is simplified from the basic part database. The analysis model generation system according to claim 1, further comprising a selection unit.   3. The analysis model generation system according to claim 1, wherein each time shape recognition of the divided model is performed, a shape corresponding to the shape recognition is stored as the basic part in the basic part database. . A shape model storing step for storing a shape model which is an analysis object of a 2D surface model or a 3D shell model created by the shape model input device;
Basic part storage process for storing basic parts with simplified basic shape models,
A shape model processing step that divides the shape model into rectangles and simplifies using basic parts of the basic shape model according to the shape recognition of each divided model divided into rectangles;
And a mesh division processing step of obtaining a mesh model which is an analysis model by dividing the simplified individual division models into meshes and combining them. A shape model rectangle dividing step of dividing the shape model into a plurality of blocks by dividing the shape model;
A divided shape recognition step for recognizing the shape of the divided model of each block divided into rectangles;
A basic part that simplifies the shape of the divided model of each block by selecting the basic part that is closest to the shape of the divided model of each block whose shape has been recognized and whose shape is simplified from the basic part database. The analysis model generation method according to claim 4, further comprising: a selection step. 6. The analysis model generation method according to claim 4, wherein a shape corresponding to the shape recognition is stored as the basic part every time shape recognition of the divided model is performed.

Images