JP2002149717A - Structure optimizing method and recording medium recorded with structure optimizing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten a design time by facilitating structure optimization as to a structure optimizing method for a design system. SOLUTION: A model generated by CAD is divided (c) into meshes, a characteristic value analysis of node coordinate vectors of the model is taken (d), and a characteristic vector is calculated. A design variable determining a candiate for an alteration of a shape is automatically generated (e) from the calculated characteristic vector and the optimization of the structure is analyzed (f) by using the generated design variable. The result of the structure optimization is outputted (g).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to a CAD (Computer).
Aided Design) system and CAE (ComputerAided)
More particularly, the present invention relates to a structure optimization method and a structure optimization program recording medium in a design system that optimizes a structure by correcting a model shape.

[0002]

2. Description of the Related Art Structural analysis systems represented by the finite element method are used in various fields in the manufacturing industry. In recent years, with the aim of shortening the design period and reducing design costs,
The introduction in the design department is also increasing. Structural optimization in a structural analysis system is a technology that combines structural analysis and numerical optimization to improve the design using a computer.

FIGS. 5 and 6 are diagrams for explaining a conventional structure optimization technique. In the structure optimization shown in FIG. 5, first, a design is performed by CAD modeling (step S1).
0). Here, optimization of the shape and the like is performed by a person based on the experience and intuition of the designer, and is directly reflected in CAD modeling. When the model is determined, a structural analysis by CAE is performed to check whether various design conditions such as a vibration problem and fatigue fracture are satisfied (step S11). If the design conditions are not satisfied, the process returns to step S10, and the design is changed based on the experience and intuition of the designer.

[0004] As a result of structural analysis by CAE, if it is confirmed that various design conditions are satisfied, a product is actually manufactured as a prototype, and a vibration test, a fatigue fracture test and other various experiments are performed (step S12). If an unsatisfactory result is obtained, the process returns to step S10 to repeat from CAD modeling.

According to this method, the number of repetitions by trial and error is large, and it takes an order of several months from the start to the end of the design.

FIG. 6 shows a conventional technique which improves this design method and automates the structure optimization as much as possible.
In this method, first, a design by CAD modeling (step S20) is performed, and then, based on an input from a designer,
Various structural analysis conditions such as physical properties, constraints, and loads of the material are set (step S21), and the model is converted into a predetermined small element mesh (VOXEL) in order to perform the structural analysis using the finite element method. It is divided (step S22).

Next, the designer examines conditions (design constraints) for performing structural optimization such as shape change candidates and the like, and design variables which are the objectives (objective functions) of structural optimization (step S2).
3), these optimization conditions are set (step S2)
4) Structural optimization is performed by structural optimization analysis using the finite element method (step S25). If the desired result is not obtained, the process returns to step S23 to re-examine the design variables, returns to step S21, changes the material and other structural analysis conditions, and obtains a satisfactory result. If not, a process of repeating redesign by CAD modeling from step S20 is performed.

According to the method shown in FIG. 6, it takes, for example, about several weeks from the start to the end of the design.

In the method shown in FIG.
As a technique for simplifying a condition input operation in setting optimization conditions of 23, "Design System and Recording Medium" described in JP-A-2000-268197
There is.

This is because when a designer / analysis engineer defines a basis vector which is a candidate for changing the shape of a model,
The edge of the model displayed on the display is picked by a mouse or the like, and the edge is deformed by dragging the picked edge, and the shape is optimized using the basis vector method as a candidate for shape change. As a result, conventionally, when inputting shape change candidates, the coordinate values of many nodes in the mesh-divided model have been defined one by one. It is now possible to enter. However, this method is basically based on the design variable (candidate for shape change)
Was the same, and it involved trial and error.

[0011]

In the conventional method shown in FIG. 5, the design was optimized by a human.
In the second conventional method shown in FIG.
25) is automatically performed by simulating with a computer. However, even in the second conventional method, design variables must be defined by a designer (or an analysis engineer) based on his / her own experience and intuition, and thus structural optimization depends on human skills. Therefore, it was necessary to derive an optimal structure by changing design variables by trial and error.

The present invention solves the above-mentioned problems, and in the optimization of a shape design for a vibration problem or the like, a structure optimization can be easily performed by automatically generating design variables, and a trial is performed. It is an object of the present invention to reduce design time by eliminating the need for repetition by mistake.

[0013]

SUMMARY OF THE INVENTION The present invention relates to a design system for optimizing a structure by correcting the shape of a model.
The model created by CAD is divided into meshes, and eigenvalue analysis is performed on node coordinate vectors related to elements of the model to calculate eigenvectors. A design variable for determining a shape change candidate is automatically generated from the calculated eigenvector, and a structural optimization analysis is performed using the generated design variable. The result of the structure optimization is output.

In the generation of design variables, eigenvalue analysis is performed on the mesh-divided model, and a plurality of low-order eigenvectors are determined as basis vectors to define design variables, and a structural variable is formed using the basis vector method. Perform optimization.

A program for realizing each of the above-described processing means by a computer can be stored in an appropriate recording medium such as a computer-readable portable medium memory, a semiconductor memory, and a hard disk.

The operation of the present invention is as follows. For the CAD data, the designer defines conditions (materials, constraints, loads) used in the structural analysis, design constraints for structural optimization, and an objective (objective function) for structural optimization. If the designer does only this, the system performs automatic mesh division without providing design variables that are candidates for shape change from the outside, and performs structural analysis using the finite element method (including eigenvalue analysis).
And optimizes the structure automatically. Therefore, the structure can be optimized easily and quickly.

[0017]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The outline of a specific procedure for implementing automatic structure optimization according to the present invention is as follows. First, (1) to (3) are executed as preparation processing, and then (4)
(8) is performed automatically by the system. (1) Model the object to be optimized using CAD (initial design). (2) Set conditions such as materials, loads, and constraints for structural analysis. (3) Set conditions such as objective functions and design constraints for structural optimization. (4) Divide the model into meshes (using existing technology). (5) Extract eigenvectors by eigenvalue analysis (using existing technology). (6) The eigenvector is used as a basis vector (candidate for shape change), and design variables are automatically generated. (7) Set design variables for structural optimization. (8) Perform a structural optimization analysis and search for an optimal solution with the maximum or minimum value of the objective function (using existing technology).

The present invention differs from the prior art in that the above-mentioned procedures (5) to (7) are performed automatically by the system.

FIG. 1 shows an example of the configuration of a structure optimization processing apparatus embodying the present invention. The structure optimization processing device 1 has a CPU
And a computer including a memory and the like, and a data creation processing unit 10, a structure optimization calculation processing unit 20, and a result evaluation processing unit 3 configured by a software program or the like.
0, a control unit 40 is provided. In the structure optimization processing device 1,
An external storage device, a keyboard and other input devices 2 and a display device 3 are connected.

The data creation processing unit 10 is a preprocessor for preparing for optimization of the structure.
An initial shape reading / display unit 11 for reading CAD data of a model created using D, a condition setting unit 12 for setting conditions for structural analysis, and a finite element dividing unit 13 for dividing the model into meshes are provided.

The structure optimization calculation processing unit 20 includes a structure analysis unit 21 including an eigenvalue analysis unit 211, a design sensitivity analysis unit 22,
A numerical optimization unit 23 is provided. The structural analysis unit 21 performs a structural analysis using the finite element method (FEM). In particular, the eigenvalue analysis unit 211 performs an eigenvalue analysis on a node coordinate vector related to an element of the model, and extracts an eigenvector. The design sensitivity analysis unit 22 is a processing unit that quantitatively calculates the degree of influence of the change of the design variable on the objective function and the design constraint (this is called a sensitivity coefficient). Numerical optimization unit 2
Reference numeral 3 denotes a processing unit for exploring and finding an optimal solution from the sensitivity coefficient and the value of the design constraint.

The result evaluation processing unit 30 edits and outputs the processing result of the structure optimization calculation processing unit 20, and includes a result editing / processing unit 31 and a result output / display unit 32. The control unit 40 controls the entire control function.

FIG. 2 is a processing flowchart according to the embodiment of the present invention. Hereinafter, (a) to (g) shown in FIG.
It will be described according to.

(A) Read / display of initial shape The initial shape read / display unit 11 of the data creation processing unit 10
CA of the model of the optimization target created using CAD
The D data 51 is input from the input device 2 configured by an external storage device such as a magnetic disk. The model shape of the CAD data 51 read as necessary is displayed on the display device 3.

(B) Setting and Display of Conditions The condition setting section 12 sets conditions 53 such as physical properties, loads and constraints of the material for structural analysis. In addition, conditions such as an objective function for structural optimization and design constraints are set. The objective function is a function for determining whether it is an optimal solution, and maximizing or minimizing the value of the objective function is an object of optimization. The design constraint is defined by giving an upper or lower limit to the results of structural analysis (displacement, stress, weight, etc.). These set conditions are displayed on the display device 3, and when the designer confirms them, the information is stored as condition data 54, and the process proceeds to the next process.

(C) Finite element division (mesh division) The finite element division unit 13 divides a model to be subjected to structural optimization based on the shape data 52 and the condition data 54 obtained from the CAD data 51 by the finite element method. For analysis using (FEM), mesh of small elements (VOXE)
L). This result is referred to as FEM data 55.

(D) Eigenvalue analysis The eigenvalue analysis unit 211 of the structure optimization calculation processing unit 20
The EM data 55 is input, an eigenvalue analysis is performed on a total synthesis matrix (node coordinate vector) including node data of elements constituting the model, and an eigenvector 56 is extracted. The method of extracting eigenvectors by the eigenvalue analysis is an existing technique and is well known, so that the detailed description of the extraction method itself will be omitted.

(E) Generation of Design Variables Next, the structure optimization calculation processing unit 20
A design variable is defined as a basis vector using a plurality of eigenvectors calculated from the low-order eigenvectors as basis vectors. The eigenvector may be considered to indicate how the original shape changes due to vibration. Since the basis vector determines a so-called deformation pattern, it is considered to be suitable as, for example, a candidate for an optimum shape in a vibration problem by defining the basis vector using an eigenvector. Defined design variable data 57
Is used for the next structural optimization analysis.

(F) Structural optimization analysis The structural optimization calculation processing unit 20 includes the condition data 54, the FEM
The data 55 and the design variable data 57 are input, and the structure analysis unit 21, the design sensitivity analysis unit 22, and the numerical optimization unit 23 perform a structure optimization calculation. First, structural analysis and design sensitivity analysis are performed on the initial shape using the finite element method. Next, in the numerical optimization, an optimal solution is exploratively obtained from the sensitivity coefficient and the design constraint value using an objective function. To reduce the processing time, a numerical model called an approximate model is used. In this method, an objective function and design constraints are calculated using an approximate expression based on Taylor expansion. The above processing
While changing the shape of the model according to the design variables, the process is repeated many times until an optimal solution satisfying the design conditions is obtained.

The processing of the structure optimization analysis is an existing technique, and for example, a technique described in the following reference can be used. [Reference 1] Vanderplaats, GN: Numerical Optimization Techniqu
es for Engineering Design: With Applcations. McGraw
-Hill, 1984. [Reference 2] Suzuki, Yamaura, Akahori: Development and application of a structural optimization system:
Magazine FUJITSU, Vol.48, No.1, 1997-1. (Http://www.fujitsu.co.jp/hypertext/develop/magaz
ine / vol48-1 / 7.html) (g) Output and display of optimization result If the result data 58 is obtained by the structural optimization analysis, the result evaluation processing unit 30 The result data 58 by the result editing / processing unit 31 in response to a request
Is edited and processed, the edited result data 59 is displayed on the display device 3 by the result output / display unit 32, and the result data 59 is output and stored in an external storage device such as a magnetic disk.

[0031]

DESCRIPTION OF THE PREFERRED EMBODIMENTS A description will be given of an application example of optimization to a disk arm shape. FIG. 3A shows an initial design shape of a disk arm designed by CAD. Here, shape optimization for maximizing the natural frequency of the in-plane vibration mode of the disk arm is attempted. The objective function is to maximize the fourth-order mode of the in-plane vibration. The design constraint is to keep the overall weight below the initial value.

As a design variable, a low-order three mode (eigenvector) of in-plane vibration is defined as a basis vector, and an arm shape is defined by the basis vector. FIG. 4 shows the three design variables 1,
Examples 2 and 3 will be described. For example, Y = ( _yx1 , _yy1 , _yz1 , _yx2 , _yy2 , _yz2 ,..., Y
_xN , _yyN , _yzN ) is a vector including the node coordinates of the entire model. When this node coordinate vector is controlled by ndv design variables, a new node coordinate vector Y _new when the design variable changes is given by: Y _new = Y ₀ + X ₁ · φ ₁ + X ₂ · φ ₂ + ... + X _ndv
Is given by φ _ndv Here, Y ₀ is the initial value of the node coordinate vector. φ _i is the eigenvector and X _i is the design variable after the change.

FIG. 3B shows the disk arm shape as a result of the automatic structure optimization according to the present invention. Conventionally,
The present invention has reduced the design time from several hours to several days, compared to several weeks of design time when humans considered and set the design variables.

[0034]

As described above, according to the present invention,
The optimization of the structure can be realized only by computer simulation, without trial and error of the designer and the analysis engineer. In particular, since the setting of the design variables does not depend on the skill of the designer, even an inexperienced young designer can derive the optimum shape. Also, after setting the necessary conditions, complete automation is possible, and the design and development time can be significantly reduced. In particular, for vibration problems, etc., design variables are defined using eigenvectors.
A better optimal solution can be obtained.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a configuration example of a structure optimization processing apparatus that implements the present invention.

FIG. 2 is a processing flowchart according to the embodiment of the present invention.

FIG. 3 is a diagram showing a shape of a disk arm for explaining an embodiment of the present invention.

FIG. 4 is a diagram showing an example of design variables for explaining an embodiment of the present invention.

FIG. 5 is a diagram illustrating a conventional structure optimization technique.

FIG. 6 is a diagram illustrating a conventional structure optimization technique.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 structure optimization processing device 2 input device 3 display device 10 data creation processing unit (preprocessor) 11 initial shape reading / display unit 12 condition setting unit 13 finite element division unit 20 structure optimization calculation processing unit 21 structure analysis unit 211 eigenvalue analysis Unit 22 Design sensitivity analysis unit 23 Numerical optimization unit 30 Result evaluation processing unit (post-processor) 31 Result editing / processing unit 32 Result output / display unit

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Kazuya Yamaura 1403-3 Nabeyada, Tsuruga, Nagano-shi, Nagano Prefecture Inside Fujitsu Nagano System Engineering Co., Ltd. 1403 No. 3 F-term in Fujitsu Nagano System Engineering Co., Ltd. F-term (reference) 5B046 AA07 JA08

Claims (4)

[Claims] 1. A structure optimization method in a design system for optimizing a structure by correcting a shape of a model, the method comprising: dividing a model into elements having a predetermined size;
Performing eigenvalue analysis on the nodal coordinate vectors related to the model elements, calculating an eigenvector, generating a design variable that determines a candidate for a shape change from the calculated eigenvector, and constructing a structure using the generated design variable. A structure optimization method, comprising: a step of performing optimization analysis; and a step of outputting a result of the structure optimization. 2. In the step of generating the design variables, a plurality of low-order eigenvectors obtained from the eigenvalue analysis are defined as basis vectors to define the design variables.
2. The structure optimization method according to claim 1, wherein in the step of analyzing the structure optimization, the optimization is performed using a basis vector method. 3. A recording medium storing a program for optimizing a structure in a design system for optimizing a structure by correcting a shape of the model, wherein the model is divided into elements having a predetermined size. Processing, performing eigenvalue analysis on nodal coordinate vectors related to the elements of the model to calculate an eigenvector, generating design variables that determine candidates for shape changes from the calculated eigenvectors, and generating the design variables. A structure optimization program recording medium characterized by recording a program for causing a computer to execute a process of analyzing a structure optimization using the same and a process of outputting a result of the structure optimization. 4. The structure optimization program recording medium according to claim 3, wherein, in the processing of generating the design variables, a plurality of eigenvectors obtained by eigenvalue analysis from low-order eigenvectors are used as basis vectors. A structure optimization program recording medium, characterized in that in a process of defining and analyzing the structure optimization, optimization is performed using a basis vector method.