JP2001297118A - Method and device for optimizing structure - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a structure optimizing device capable of easily preparing an accurate analytical model also by a person having no expert knowledge even when the model has a complicated boundary shape and performing optimization/partial optimization followed by a topological change due to the addition/deletion of ribs, bosses or the like. SOLUTION: The structure optimizing device is provided with a means 101 for selecting an area to be optimized and followed by a topological change and defining a feature, an attribute, a standard, and an optimizing condition, a means 102 for dividing the optimized area into plural small areas, a means 103 for preparing feature definition and a three-dimensional shape, and a means 104 for preparing an analytical model and meshing a calculation area. The device is provided also with a numerical analysis means 105, a means 106 for instructing re-calculation in accordance with the evaluation results of an objective function and a restricting condition, and a means 107 for extracting the combination of feature standards so that the restricting condition is satisfied and the objective function is minimized. Consequently the inserting position of a rib or a boss of a structure can be quickly determined only by specifying an optimized position and partial optimization can be performed by using an analytical model having an accurate shape.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a CAE (Computer)
The present invention relates to a structure optimization method and a structure optimization device in Aided Engineering, and particularly to a means for optimizing a shape of a structure accompanied by a topology change due to addition / deletion of a rib, a boss, and the like.

[0002]

2. Description of the Related Art Changing and optimizing the shape of a product to increase its rigidity and reduce its weight is one of design issues. As an example of the conventional optimization method, there is an optimization based on experience and intuition of a skilled designer. In this optimization method, a skilled designer changes the design of the optimization target model, evaluates the performance such as strength and weight by structural analysis and experiments of the changed model,
Based on the evaluation result, the procedure of further changing the design is repeated to determine the optimum shape.

At this time, a skilled designer considers not only the size change of the model to be optimized, but also the topology change of inserting / deleting ribs (reinforcing members) and bosses (projections) in the structure. Change the design. An optimization method using a homogenization method has been proposed as a shape optimization method that takes into account topological changes such as insertion and deletion of ribs and bosses (KS
uzuki and N. Kikuchi, A homogenization method for sh
ape and topology optimization Comput.Methods Appl.
Mech. Engrg. 93 (1991) 291-318).

An outline of the homogenization method will be described. First, the optimization region with the topology change is divided into regions of an orthogonal lattice called voxels. Each divided small area is assumed to be a porous material. An apparently infinite number of micropores is used to calculate an elastic tensor in a state where holes are made in a small region by a homogenization method. The stiffness matrix of the entire analysis model is calculated using the calculated elastic tensor, and the structural analysis is performed by the finite element method. At this time, if the volume of the hole is equal to the volume of the small region, it means that no substance exists in the small region.
By maximizing the number of micropores, that is, the vacant portions in each small region under given constraints, an optimized shape can be determined by finding a portion that is unnecessary in terms of strength.

[0005]

The above prior art has the following problems. First, to train skilled designers,
It requires enormous amount of time and cost, and with the complexity of the structure, it is no longer possible to respond only by the experience and intuition of the designer. Optimization techniques that involve topology changes, such as the homogenization method,
Since the region to be optimized is all divided into orthogonal grids called voxels, a model shape having a complicated boundary shape cannot be accurately reflected on the analysis model of the homogenization method, and the analysis accuracy decreases. When applying the homogenization method to a part of the structure, the optimization region is divided into voxels. If the boundary between the region to be optimized and the region not to be optimized is complicated, a difference occurs. Optimization is difficult.

An object of the present invention is to easily create an accurate analysis model even for a model having a complicated boundary shape,
An object of the present invention is to provide a structure optimization method and a structure optimization device capable of performing a partial optimization with a topology change due to addition / deletion of a rib or a boss.

[0007]

According to the present invention, in order to achieve the above object, an optimization target area is input, and at least a part of the optimization target area is deleted or added to optimize a structure with a topology change. In the structural optimization method for obtaining a shape, an optimization region that allows a change in topology is selected, features to be deleted or added, the depth and level of the feature are defined, optimization conditions are defined, and the selected optimization region is defined. Is divided into a plurality of small areas, a three-dimensional shape is created for each of the divided small areas based on the definition of the features and their attributes, an analysis model is created from the three-dimensional shape, and boundary conditions and calculation conditions are defined. , Generate a mesh of the created analysis model, numerically analyze the entire analysis model, evaluate the objective function and constraints, terminate if the objective function is the minimum, Commands the recalculation of Udenakereba numerical analysis, the objective function is to propose a structure optimization method for extracting a combination of levels of features so as to minimize while satisfying the constraint condition.

The present invention also provides a structure optimization method for inputting a region to be optimized and obtaining an optimum shape of a structure having a topology change by deleting or adding at least a part of the region to be optimized. Select optimization areas to allow, define features to be removed or added, feature depth and level, define optimization conditions,
The selected optimization area is divided into a plurality of sub-areas, a combination of levels for each feature is generated from a certain initial value of the defined feature using a generation function, and the feature and its Create a 3D shape based on the definition of attributes, create an analysis model from the 3D shape, define boundary conditions and calculation conditions, generate a mesh of the created analysis model, numerically analyze the entire analysis model, Evaluate the objective function and constraints, terminate if the objective function is the minimum, otherwise extract a new combination using the generator function, command recalculation of numerical analysis, and Evaluate the combination calculated using the acceptance function based on the reference value, reduce the reference value as the evaluation is achieved, and minimize the objective function while satisfying the constraints. We suggest structural optimization method for extracting a combination of levels of features as.

In order to achieve the above object, the present invention further provides a structure optimization method for inputting a region to be optimized and obtaining an optimum shape of a structure having a topology change by deleting or adding at least a part of the region to be optimized. An optimization calculation input unit for selecting an optimization region allowing a topology change, defining a deleted or added feature, a depth and a level of the feature, and defining an optimization condition, and an optimization calculation input unit for defining the optimization region. An area division unit that divides a plurality of small areas, a three-dimensional shape / feature creation unit that creates a three-dimensional shape based on the definition of features and their attributes for each of the divided small areas, and an analysis model based on the three-dimensional shape. An analysis model creation unit that creates and defines the boundary conditions and calculation conditions to generate a mesh of the created analysis model, and numerical analysis of the entire analysis model An analysis unit that evaluates the objective function and constraints, and an evaluation unit that instructs recalculation of numerical analysis if the objective function is not the minimum if the objective function is not minimized, so that the objective function is minimized while satisfying the constraints A structure optimization device provided with a level selection unit for extracting a combination of feature levels is proposed.

In this structure optimization apparatus, an optimization calculation input unit, a region division unit, a feature creation unit, a three-dimensional shape creation unit, an analysis model creation unit, an analysis unit, an evaluation unit, and a level selection unit are arbitrarily combined. , A plurality of computers selected from a personal computer, a workstation, and a parallel computer.

[0011]

FIG. 1 is a system diagram showing the overall configuration of an embodiment of a structure optimizing apparatus according to the present invention. FIG.
The structure optimization apparatus includes an optimization calculation input unit 101, a region division unit 102, and a three-dimensional shape / feature creation unit 103.
, An analysis model creation unit 104, an analysis unit 105, an evaluation unit 106, a level selection unit 107, and a computer.
This computer performs a server function and a function of controlling the overall operation.

The optimization calculation input unit 101 selects a region to be optimized with a topology change, that is, an optimization region,
Define extrude or cut features,
Objective function definition, constraint condition definition,
Enter the maximum number of calculation steps.

The extruded feature is the area A in FIG.
Is a procedure for extruding a part and adding a shape to create a new overall shape. The cut feature is a procedure for deleting a part of the areas B1 and B2 in FIG. It is. In this specification, an extruded feature and a cut feature are collectively called a feature.

Also, parameters such as the depth of the feature and the number of levels are introduced. The feature depth means an extruding distance when an extruded feature is selected, and a cutting distance when a cut feature is selected. The number of levels represents the number of conditions taken up for a feature for calculation, and the conditions are called levels.

Taking an example of a measurement experiment of the amount of oil evaporation at 100 degrees, 200 degrees, and 300 degrees, the levels are 100 degrees, 200 degrees, and 300 degrees representing the experimental conditions, and the number of levels is taken up. The number of experiments is 3.

The area dividing section 102 is an optimization calculation input section 1
The optimization area defined by 01 is divided into a plurality of arbitrary areas.

The three-dimensional shape / feature creation unit 103
This is the part that creates the three-dimensional shape to be optimized.
It is generally called three-dimensional CAD (Computer Aided Design). The three-dimensional shape / feature creation unit 103 defines each area divided by the area division unit 102 as an extruded feature or cut feature defined by the optimization calculation input unit 101.

The analysis model creation unit 104 creates an analysis model, and defines physical property values, boundary conditions, and calculation conditions for structural analysis, thermal fluid analysis, and the like. In addition, a mesh of the created analysis model is generated.

The analysis unit 105 executes a numerical simulation such as a structural analysis or a thermal fluid analysis.

The evaluation section 106 evaluates the objective function and the constraint condition, and instructs recalculation based on the evaluation result and repeats the calculation.

The level selection unit 107 is provided with an optimization calculation input unit 1
The combination of the levels of the features defined for each small area is determined so as to minimize the objective function while satisfying the constraint conditions defined in 01.

FIG. 3 is a flowchart showing an example of a processing procedure in the structure optimizing apparatus shown in FIG. 1, FIG. 4 is a perspective view showing an example of a model to be optimized, and FIG. FIG. 3 is a diagram illustrating an example of an input screen configuration of a unit 101. It is assumed that a shaded portion of the three-dimensional model A is fixed to a rigid body, and a load P acts on points A and B.

In step 1, the optimization calculation input unit 1
In step 01, an optimization area is selected, features are defined, and an objective function, constraints, and the maximum number of calculation steps are input.

As shown in FIG. 5, the model of FIG. 4 is displayed on a screen for selecting an optimization area. The user selects a region to be optimized with a change in topology. In FIG. 5, the shaded area is the optimization area defined by the user. Next, features are defined. Here, there are an extruded feature and a cut feature, and at least one is selected. Enter the feature depth and number of levels. Here, the depth is 20 and the number of levels is 2.

The level number N is a natural number taking 1 ≦ N,
It is a level related to depth. If the number of levels is 1, the numerical value defined by the depth is taken as the level, and if the depth is 20, the level is 20. In the case of the number of levels 2, the numerical value defined by the level 0 and the depth is taken as the level.
The levels are 0 and 20. Generally, for the number of levels 2 ≦ N, the level is (input depth) * i / (N−1) (i = 0... N−1).

In the definition of the objective function, in this case, the mass is displayed, and the mass is selected. Further, there is no limitation on the number of objective functions, and a plurality of objective functions can be selected.

In the constraint condition definition, in this case, strain and stress are displayed, and numerical values are input to each of them. At this time, taking strain as an example of a constraint, the maximum strain of the analysis result may be used as a constraint, but an arbitrary point of the model to be optimized is input on the screen, and the Can be taken as a constraint.

The input of the number of optimum calculations specifies the maximum number of iterations in the optimization calculation.

FIG. 6 is a diagram showing an example of area division.
The region dividing unit 102 divides the optimization region selected by the optimization calculation input unit 101 into a plurality of small regions. In FIG. 6, the hatched portion selected as the optimization area in FIG. 5 is divided into a grid.

Although the area is divided into a lattice here, it may be divided into arbitrary partial areas. In addition, the small areas may overlap each other.

In the feature definition 103A, the features defined by the optimization calculation input unit 101 are defined for each small area divided by the area division unit 102. In the small area shown in FIG. 7, if an extruded feature is defined in a shaded small area and a level 20 is selected,
The small area is extruded by a height of 20.

The feature definition performed by the feature definition 103A is executed for all the small areas divided by the area dividing unit 102, and is stored as a table A as shown in FIG.

In FIG. 8, the feature name is a feature name corresponding to each small area, and the level is a depth taken up for calculating a feature defined for each small area. In Extrusion 1 of this example, two calculations of an extrusion distance 0 and an extrusion distance 20 are taken.
Similarly, for all other features in Table A, features and levels also correspond.

In step 2, an optimal calculation is performed. In the initialization process 107, a level is randomly selected one by one from each feature of the table A created in step 1. The three-dimensional shape / feature creation unit 103B creates a three-dimensional shape according to each feature and the selected level.

The analysis model creation unit 104 creates an analysis model, defines boundary conditions, and defines calculation conditions. Here, the boundary conditions shown in FIG. 4 are given. Further, a mesh of the analysis model is generated.

The analysis unit 105 analyzes the structure of the created analysis model under given boundary conditions and analysis conditions.

The evaluator 106 evaluates the objective function. Satisfy the constraints defined by the optimization calculation input unit 101,
If the objective function is the minimum, the shape model in that step is stored as an optimal solution.

Otherwise, the level selection unit 107 extracts a combination of the levels of each feature from the table A by using an optimization method so as to reduce the objective function while satisfying the constraint conditions, The calculation is performed again from the procedure of the creation unit 103B.

When the number of calculation steps is larger than the maximum number of calculation steps, the result of the step with the smallest objective function value while satisfying the constraint conditions in each iterative calculation is determined as the optimal solution of the optimum shape, and the shape model is changed. save.

As for the optimization method, various methods can be selected. In this case, the simulated annealing (Sim
An example in which the ulated annialing method is applied as an optimization method will be described. FIG. 9 is a flowchart showing a processing procedure when the simulated annealing method is used.

Optimization calculation input unit 101, region division unit 10
2, three-dimensional shape creation unit 103, analysis model creation unit 10
4. The processing procedure in the analysis unit 105 is the same as that in the embodiment of FIG.

In the simulated annealing method, a new value is generated from a certain initial value by using a generating function. Here, a combination of levels for each feature is generated.
Perform calculations on the generated values. Next, based on a certain reference value (here, a reference value for the objective function), the result calculated by using the reception function is evaluated, and it is determined whether or not to accept the originally generated value. If accepted, the generated value is used as an initial value, and a new function is used to generate a new value. This process of generation and acceptance is repeated until an equilibrium state is reached, and when the equilibrium state is reached, the next reference value is obtained using the slow cooling function (here, the objective function is reduced). When the reference value reaches the stop condition (minimum), the calculation is terminated as an optimal solution.

In the initialization process 107, one level is randomly selected from each level of the table A created in step 1 and is set as an initial value (combination A), and a new combination (combination B) is generated by using a generating function. To determine. The three-dimensional shape / feature creation unit 103B creates a three-dimensional shape according to the feature combination B generated by the generation function. The analysis model creation unit 104 creates an analysis model, defines boundary conditions and calculation conditions, and generates a mesh. Analysis unit 10
At 5, the structure is analyzed. When the objective function reaches the stop condition (minimum), the evaluation unit 106 stores the shape model data at that time as an optimal solution.

Otherwise, the optimization calculation input unit 101
With reference to the objective function and constraints defined in, use the acceptance function for the reference value corresponding to the set objective function,
Determine whether to accept the calculated value. If not accepted, the level selection unit 107 extracts a new level combination of each feature using a generation function based on the combination A, and
The calculation is performed again by the dimensional shape creation unit 103B. When accepting, the level selection unit 107 extracts a new combination using the generation function with the combination B as an initial value, and causes the three-dimensional shape creation unit 103B to recalculate the combination. Iterative calculation is repeated until the value generated by the generating function reaches an equilibrium state.

When an equilibrium state is reached, the reference value is lowered using a slow cooling function. Thereafter, if the number of iteration calculations exceeds the maximum number of step calculations, the calculation is terminated, and the result when the objective function is the smallest while satisfying the constraints in each iteration calculation is taken as the optimal solution, and the shape model data at that time is used. Save.

FIG. 10 is a diagram showing an example of an optimization calculation involving a topology change obtained by the above procedure. A rib shape is generated in the optimization area, and the position is accurately recognized. As a result, it is possible to obtain a rib insertion position that satisfies the required constraints while reducing the mass.

FIG. 10 shows an optimal result obtained when the optimized region is arbitrarily divided into regions. On the other hand, if the design intention such as the rib shape is clear in advance, FIG.
As shown in (1), when the region is divided in consideration of the design intention and the structure optimization method of the present invention is applied, the rib insertion position can be determined more accurately.

Here, the personal computer is assigned to the interactive part, that is, the function of step 1 in FIG. 3 or FIG. 9, and another personal computer, workstation, parallel computer, etc. is assigned to the function of step 2. Although a form was assumed, the present invention is not limited to this allocation example. A plurality of computers selected from a personal computer, a workstation, and a parallel computer by arbitrarily combining the optimization calculation input unit 101 to the level selection unit 107 in FIG. 1 so as to obtain a structure optimization device with good work efficiency and quick response. The present invention is also effective in a case where the components are mounted separately.

[0049]

According to the present invention, an optimum shape of a structure can be obtained only by inputting an optimization area, so that a structure involving a topology change due to addition / deletion of ribs, bosses, etc. can be optimized in a short time. it can. Further, since the shape created by the three-dimensional shape / feature creation unit is used as the analysis model, even for a structure having a complicated boundary, the boundary can be faithfully reflected on the analysis model, and a decrease in analysis accuracy can be prevented. Furthermore, when optimizing a part of the structure, partial optimization or optimization is performed by executing only partial deletion or addition of the optimization area, and thus optimization is possible.

[Brief description of the drawings]

FIG. 1 is a system diagram showing an overall configuration of a structure optimization device according to the present invention.

FIG. 2 is a diagram illustrating an extrusion feature and a cut feature.

FIG. 3 is a flowchart illustrating an example of a processing procedure in the structure optimization device of FIG. 1;

FIG. 4 is a perspective view showing an example of an optimization target model.

FIG. 5 is a diagram illustrating an example of a configuration of an optimization calculation input screen.

FIG. 6 is a diagram illustrating an example of area division.

FIG. 7 is a diagram illustrating an extruded feature executed in a small area.

FIG. 8 is a chart showing an example of a table describing each feature and its level.

FIG. 9 is a flowchart showing a processing procedure when a simulated annealing method is applied to an optimization method.

FIG. 10 is a diagram illustrating an example of an optimization calculation involving a topology change due to addition of a rib, a boss, and the like.

FIG. 11 is a diagram illustrating an example of an optimal calculation result when a region is divided in consideration of a design intention.

[Explanation of symbols]

 101 optimization calculation input unit 102 area division unit 103 three-dimensional shape / feature creation unit 104 analysis model creation unit 105 analysis unit 106 evaluation unit 107 level selection unit

 ────────────────────────────────────────────────── ─── Continuing on the front page (72) Inventor Makoto Onodera 502 Kandate-cho, Tsuchiura-shi, Ibaraki F-term in the Machine Research Laboratory, Hitachi Ltd. 5B046 DA02 FA18 GA01 JA08

Claims (4)

[Claims] 1. A structure optimization method for inputting a region to be optimized and obtaining an optimum shape of a structure having a topology change by deleting or adding at least a part of the region to be optimized. Selecting the optimization area, defining the depth or level of the deleted or added features and the features, defining optimization conditions, dividing the selected optimization area into a plurality of small areas, and dividing each of the divided areas. A three-dimensional shape is created based on the definition of the features and their attributes for the small region of the above, an analysis model is created from the three-dimensional shape, boundary conditions and calculation conditions are defined, and a mesh of the created analysis model is generated. Numerical analysis of the entire analysis model, evaluating the objective function and constraints, terminating if the objective function is the minimum, otherwise, A structural optimization method characterized by instructing recalculation and extracting a combination of feature levels so as to minimize an objective function while satisfying constraints. 2. A structure optimization method for inputting a region to be optimized and obtaining an optimum shape of a structure having a topology change by deleting or adding at least a part of the region to be optimized. Selecting the optimization area, defining the depth or level of the feature to be deleted or added and the feature, defining optimization conditions, dividing the selected optimization area into a plurality of sub-areas, defining the defined feature Generating a combination of levels for each feature from a certain initial value using a generation function, creating a three-dimensional shape based on the definition of the feature and its attributes for each of the divided small regions, and analyzing from the three-dimensional shape Create a model, define boundary conditions and calculation conditions, generate a mesh of the created analysis model, Is numerically analyzed, the objective function and the constraint conditions are evaluated, and if the objective function is the minimum, the process is terminated.If not, a new combination is extracted using the generation function, and a recalculation of the numerical analysis is instructed. Evaluate the combination calculated using the acceptance function according to the reference value for the objective function, and reduce the reference value as the evaluation is achieved, and reduce the level of the feature so that the objective function is minimized while satisfying the constraints. A structure optimization method characterized by extracting a combination. 3. A structure optimizing apparatus for inputting an optimization target region and obtaining an optimum shape of a structure having a topology change by deleting or adding at least a part of the optimization target region, wherein the optimization allows a topology change. An optimization calculation input unit for selecting an area, defining the deleted or added features and the depth and level of the features, and defining optimization conditions; and an area division for dividing the selected optimized area into a plurality of small areas. A three-dimensional shape / feature creation unit for creating a three-dimensional shape based on the definition of the features and their attributes for each of the divided small regions; a boundary condition and calculation conditions for creating an analysis model from the three-dimensional shape An analysis model creating unit that generates a mesh of the created analysis model, and an analysis unit that numerically analyzes the entire analysis model; An evaluation unit that evaluates the objective function and constraints and terminates if the objective function is minimum, otherwise recalculates the numerical analysis, and a combination of feature levels so that the objective function is minimized while satisfying the constraints And a level selecting unit for extracting a value. 4. The structure optimization device according to claim 3, wherein the optimization calculation input unit, the region division unit, the feature creation unit, the three-dimensional shape creation unit, the analysis model creation unit, and the analysis unit. And an evaluation unit and the level selection unit are arbitrarily combined, and are distributed and mounted on a plurality of computers selected from a personal computer, a workstation, and a parallel computer.