JP4374227B2 - Shape optimization processor - Google Patents

Description

  The present invention relates to a shape optimization method and apparatus for optimizing a design shape by analysis on a computer using a finite element method when designing a structure such as a mechanical device or its parts.

  For example, in designing a mechanical structure, the design shape of the mechanical structure is changed in various ways by selecting design variable values within a given design specification, and the performance of each shape is measured and measured. By analyzing the sensitivity of the performance obtained in step 1, the optimum shape that meets the design specifications is found. That is, a design solution for the optimum shape is obtained by executing a number of parameter surveys. In such design work, support for digital engineering technology represented by CAD (Computer Aided Design) that draws the shape of a structure on a computer and CAE (Computer Aided Engineering) that performs numerical analysis on the same computer. Accordingly, it is widely performed to improve the efficiency of design solution derivation. However, when using CAE, the obtained numerical analysis result on the computer must be sufficiently close to the actual result. That is, for example, when obtaining a stress value by numerical analysis, the result on the computer must be a value sufficiently close to an actual stress value in an actual product produced based on the design. For such analysis, it is necessary to improve the analysis accuracy by sufficiently finening the finite element mesh in the finite element method used in the numerical analysis in CAE.

  As a conventional technique for efficiently supporting a design using digital engineering, for example, a technique disclosed in Patent Document 1 is known. In this prior art, an orthogonal table allocating step for arranging data related to design factors and levels set based on the analysis target in columns and rows of the orthogonal table selected according to the design factors and levels, and the orthogonal table A structural analysis step for performing structural analysis on the data of each arranged row, a dispersion analysis step for performing dispersion analysis based on the result of the structural analysis step, and a characteristic value representing the property to be analyzed based on the above dispersion analysis Extraction of design factors and order components with large effects and creation of an estimation formula using orthogonal functions based on them, and the objective based on the mathematical optimization calculation method after executing this estimation formula creation step It includes an optimization design process that performs optimization calculation.

Japanese Patent Laid-Open No. 10-207926

  As described above, in the design support technology using the conventional digital engineering, the design efficiency is improved by combining the CAE and the optimization calculation technology. A parameter survey is required for the optimization calculation, and the execution involves many iterative calculations. Therefore, it takes time to multiply one calculation time by the number of iterations, and in order to obtain a reliable numerical analysis solution, it is necessary to make the finite element mesh sufficiently fine as described above. As a result, one calculation time becomes long, and a lot of calculation time is consumed.

  In other words, in the past, using an optimization calculation technique to derive an optimal shape requires a large amount of processing time to obtain a highly reliable analytical solution. There was a relationship that the reliability decreased. With respect to such a problem of processing time, in the technique of Patent Document 1, characteristic values representing performance are calculated by a parameter survey, and the estimation formulas are calculated to reduce the time for deriving the design solution. However, the technique of Patent Document 1 does not always sufficiently consider the accuracy of analysis.

The present invention has been made against the background of the above-described conventional circumstances, and an object thereof is to provide a shape optimization processing apparatus that can perform high-accuracy analysis and can significantly reduce processing time. Yes.

For the above-described purpose, in the present invention, an analysis model definition unit that sets an analysis model related to the shape of a design target product, an optimization condition definition unit that sets optimization conditions for the analysis model, and a finite number from the analysis model. Create an element mesh and perform a preliminary fine analysis by repeating the analysis with the first division density at the first finite element mesh at a first iteration number, and the second division density coarser than the first division density. An analysis unit for performing a preliminary rough analysis in which the analysis with a finite element mesh is repeated at the second iteration number different from the first iteration number or the first iteration number; and the result of the preliminary fine analysis; Based on the difference from the result of the preliminary rough analysis, a correction function creating unit for creating a correction function, and a constraint condition by correcting each result of the analysis in the finite element mesh by the second division density with the correction function Value and That the constraint value correcting section, the constraint value first number of iterations and the constraints value after corrected by a large third number of iterations than the second repetition rate by the correction unit, satisfies the optimization condition And an optimal calculation control unit for determining whether or not there is.

Further, in the present invention, for the shape optimization processing apparatus as described above, the analysis unit performs a verification analysis to analyze with a finite element mesh having a third division density smaller than the second division density, and performs the optimum calculation. The control unit determines whether or not the result of the verification analysis satisfies the optimization condition .

Further, in the present invention, for the shape optimization processing apparatus as described above, when the optimization calculation control unit determines that the optimization condition is not satisfied, an optimization condition change display unit that displays the fact on the screen It is equipped with a.

Further, in the present invention, the shape optimization processing apparatus as described above is configured so that iterative analysis in each of the preliminary fine analysis and the preliminary rough analysis is performed by simultaneous processing.

As described above, in the present invention, the analysis for deriving the design solution (the main analysis) is performed with a coarse finite element mesh, so that the processing time can be greatly shortened. On the other hand, a correction function is created by performing preliminary fine analysis and preliminary rough analysis with fewer iterations than the number of iterations in this analysis, and each correction result in this analysis is corrected with this correction function. Therefore, it is possible to ensure sufficiently reliable analysis accuracy. That is, according to the present invention, highly accurate numerical analysis can be performed, and the processing time can be greatly shortened.

FIG. 1 is a block diagram showing the overall configuration of a shape optimization processing apparatus according to an embodiment. This shape optimization processing apparatus includes a computer 110 and a database storage device 109 as hardware elements. In the computer 110, an analysis model definition unit 101, an optimization condition definition unit 102, an analysis unit 103, a correction function creation unit 104, a constraint value correction unit 105, an optimal calculation, which are formed as processing means based on a computer program, respectively. A control unit 106, a correction information output unit 107, and an optimization condition change display unit 108 are mounted.

  The analysis model definition unit 101 creates an analysis model (three-dimensional model) of a design target product that is a shape optimization target. The optimization condition definition unit 102 defines conditions for optimal calculation (optimization calculation). The optimal calculation is to find a solution that minimizes or maximizes the objective function under the constraints given as design specifications. Accordingly, the conditions in the optimum calculation include calculation conditions such as boundary conditions necessary for numerical analysis, and objective functions, design variables, and constraint conditions necessary for shape optimization. Further, the optimization condition definition unit 102 also inputs an allowable value that is a criterion for determining whether or not the model (shape as a design solution) finally obtained by the optimal calculation correctly satisfies the constraint condition.

The analysis unit 103 performs numerical analysis by creating a finite element mesh from a three-dimensional shape based on the analysis model. In the numerical analysis performed by the analysis unit 103, a preliminary analysis to create a correction function, the final analysis result that is present analysis to derive design solutions, and validation analysis for verifying the results of the analysis There is. In the preliminary analysis for creating the correction function, the preliminary fine analysis on the finite element mesh by the first division density is repeated at the first iteration number, and the finite by the second division density which is coarser than the first division density. The preliminary coarse analysis on the element mesh is repeated with a first iteration number or a second iteration number different from the first iteration number. The correction function creation unit 104 creates a correction function based on the difference between the two analysis results. In the present analysis for deriving the design solution, the analysis with the finite element mesh with the second division density is repeated at the first iteration number or the third iteration number larger than the second iteration number, and each of the iteration analysis is performed. The final analysis result is derived while correcting the result with the correction function. In the verification analysis, the shape based on the final analysis result is analyzed with a finite element mesh having a third division density smaller than the second division density, and it is determined whether or not the analysis result satisfies the design specification.

  The correction function creation unit 104 creates a correction function based on the preliminary analytical solution by the analysis unit 103 as described above. This correction function takes the form of an approximate polynomial as will be described later. The constraint condition value correcting unit 105 corrects the result of the optimal calculation. Specifically, the numerical analysis result obtained with respect to the constraint condition at the time of execution of the optimal calculation is corrected using the correction function obtained by the correction function creation unit 104, and the corrected new value is finalized. Limit value. The optimal calculation control unit 106 changes the design variable so that the objective function is minimized or maximized while satisfying the constraint condition, and re-executes the analysis if the solution does not converge, and ends the calculation if the solution converges. Control optimization calculations. The correction information output unit 107 visualizes and displays the correction function created by the correction function creation unit 104, displays the transition of the result of optimal calculation, and the result of verification analysis. The optimization condition change display unit 108 prompts the user to review the optimization condition (optimal calculation condition) when it is determined that the model obtained by executing the optimal calculation does not correctly satisfy the constraint condition in the verification analysis. Is displayed to the operator. The database stored in the storage device 109 manages information required by the present apparatus, such as analysis models, optimization conditions, and correction functions.

It described below shape optimization Kasho management executed by the configuration of the shape optimization processing apparatus as described above. Figure 2 is a flowchart illustrating the flow of shape optimization process performed by the shape optimization processing apparatus of FIG. As seen in this figure, the shape optimization process in the present invention is roughly divided into three phases. The first phase 1 is a phase for inputting an analysis model and conditions necessary for optimization calculation. The second phase 2 is a phase for creating a correction function, which is a correction function creation step. The third phase 3 is a phase for executing the optimal calculation which is the main analysis. In the optimum calculation, as described above, the analysis proceeds while correcting the numerical analysis result regarding the constraint condition obtained during the calculation using the correction function created in the phase 2.

  FIG. 3 shows an example of a machine part that is a design target product. In the following, description will be made on a phase-by-phase basis, taking as an example the case where optimal calculation is performed for the design target product. Here, although the machine part shown in FIG. 3 is a single part, it is needless to say that the present invention can be applied to an assembly including a plurality of parts.

  Phase 1 includes step S1 and step S2. First, in step S1, an analysis model to be subjected to shape optimization is created (defined). Specifically, the shape of the machine part shown in FIG. 3 is created. The analysis model is created by using the analysis model definition unit 101. The analysis model definition unit 101 normally uses a three-dimensional CAD or CAE preprocessor.

  In step S2, optimization conditions are defined (input). The optimization condition is defined using the optimization condition definition unit 102. FIG. 4 shows an example of an optimization condition input screen by the optimization condition definition unit 102. The optimization condition input screen in this example includes a screen for inputting an analysis model to be optimized for shape. Therefore, the processing in step S1 and step S2 is performed with this optimization condition input screen activated. In the definition of the optimization condition in step S2, calculation conditions, objective functions, and constraint conditions necessary for numerical analysis such as boundary conditions are performed. And conditions necessary for optimization calculations such as design variables. Also, in the definition of optimization conditions, the tolerance value, which is the criterion for determining whether the model finally obtained by the optimization calculation correctly satisfies the constraint conditions, that is, the final shape (this is the shape obtained as the design solution by the optimization calculation) Also, an allowable value that is an allowable amount for the difference between the constraint condition value obtained by correcting the result of the optimal calculation with the correction function and the constraint condition value obtained by the verification analysis is input. In the screen example of FIG. 4, the machine part of FIG. 3 to be optimized is input, boundary conditions necessary for numerical analysis are defined, and design variables for optimal calculation are also defined. Also, mass is specified as the objective function, and minimization is selected as the objective of optimization. Stress is specified as the constraint condition, and the constraint range (allowable region) is input. Specifically, −5 × 9.8 MPa is input as the lower limit value of stress, and 5 × 9.8 MPa is input as the upper limit value. In this stress value, the minus side is compressive stress and the plus side is tensile stress. In addition, “dimension A” and “angle A” of the machine part are defined as design variables, and the constraint range is input. Specifically, “dimension A” is input with a lower limit value of 10 mm, an initial value of 13 mm, and an upper limit value of 16 mm, and “angle A” with a lower limit value of 0.01 degrees, an initial value of 4 degrees, 8 degrees is input as the upper limit value. Further, 10% is input as an allowable value. This is Phase 1.

  The next phase 2 is a phase for creating a correction function, and includes steps S3 to S5. In phase 2, a preliminary analysis is performed to create a correction function. Preliminary analysis consists of a set of preliminary fine analysis and a set of preliminary coarse analysis. Therefore, a total of two sets of numerical analyzes are performed in the preliminary analysis. The preliminary fine analysis is performed with a finite element mesh having a fine first division density so as to obtain a sufficiently reliable numerical analysis result. On the other hand, the preliminary rough analysis is a numerical analysis with a large error but a short calculation time, and is performed with a finite element mesh having a second division density coarser than the first division density. In both of these analyses, the analysis is repeated with the first iteration number or the second iteration number, respectively.

  In the case of the example of FIG. 4, each of the preliminary fine analysis and the preliminary rough analysis satisfies the conditions described later from between the upper limit value and the lower limit value of each of the design variables “dimension A” and “angle A”. It is repeated while arbitrarily selecting values in the range, and the creation of a three-dimensional shape for the selected design variable value (step S3) and the execution of numerical analysis for the three-dimensional shape (step S4) are sequentially performed. It is done by repeating. Based on the result of this repeated analysis, a correction function is created in step S5.

  In the creation of the three-dimensional shape in step S3, a three-dimensional shape is created using a three-dimensional CAD or CAE preprocessor in the same manner as the creation of the analysis model. The three-dimensional shape is created based on the analysis model defined in step S1 by using a value arbitrarily selected from the upper limit value and the lower limit value of the design variables in the example of FIG.

  In the execution of numerical analysis in step S4, a preliminary fine analysis is performed by creating a fine finite element mesh with the first division density from the three-dimensional shape created in step S3, and the three-dimensional shape also created in step S3. A rough finite element mesh with the second division density is created from the above, and preliminary coarse analysis is executed.

In the creation of the correction function in step S5, a correction function is created from the result of the repeated analysis using the correction function creation unit 104. The correction function is created as an approximate polynomial having a difference Δ _i between the results of the preliminary fine analysis and the preliminary rough analysis as a dependent variable and design variables (specifically, dimension A and angle A) as independent variables. Here, the subscript i of Δ _i represents an index. The distance between the design variable values selected in the above iterative analysis is defined by X ^MAX = Max | X ⁱ −X |. If X ^MAX ≦ ε, the distance between the taken analysis results is related. Create an approximate polynomial by the method of least squares. On the other hand, if X ^MAX ≦ ε, the number of iterations of the analysis is still insufficient for the creation of the correction function, so the difference between the analysis results at that index and the design variable value are stored in the computer 110 and the correction function is created. The analysis is repeated until a new design variable value is selected. The design variable value is selected at random so that ε−e ≦ X ^MAX in the region of the constraint condition related to the design variable (in the example of FIG. 4, the upper limit value and the lower limit value of the design variable).

Here, e is an adjustment parameter, and a sufficiently small value is used for ε. That is, ε−e ≦ X ^MAX is adjusted so that the values of the selected design variables do not become too close to each other. From this, X ^MAX ≦ ε gives the lower limit of the number of iterations of preliminary fine analysis and preliminary coarse analysis, whereas ε-e ≦ X ^MAX gives the upper limit of the number of iterations of preliminary fine analysis and preliminary coarse analysis. It can be said that X ⁱ is a design variable value selected by the index, and X means all design variable values selected in the past. That is, X ^MAX means that the maximum value of each distance is taken with respect to all design variables selected in the past. Accordingly, the selection points are scattered within ε in the design variable selection region. This ε is arbitrarily set, but in a preferred example, it takes 20% of the design variable selection area. Taking dimension A as an example, since the upper limit is 16 mm and the lower limit is 10 mm, ε is 1.2 mm.

In the above, a case has been described in which the design variable values of the dimension A and the angle A are selected one by one for the example of FIG. 4 and the processing of step S3 and step S4 is repeated. Is also possible. That is, at least one or more (one or more sets) of design variable values satisfying X ^MAX ≦ ε are calculated together, and in step S3, three-dimensional shapes with the respective design variable values are collectively generated, and step S4 is performed. In this case, it is also possible to perform a numerical analysis on each of the plurality of three-dimensional shapes in parallel. When such concurrent processing is performed, a computer network is used to work in a distributed environment. In the above description, the selection of the design variable value is random. However, the value of the design variable can be determined using a method based on an experimental design method such as an orthogonal table.

  There is no particular restriction on the polynomial of the correction function, but one preferred example is to use the following second-order polynomial.

ここで、Ｘ₁,Ｘ₂はそれぞれ寸法Ａと角度Ａを表わす。またａ〜ｆは多項式の係数を意味し、これら各係数は最小２乗法を利用することにより求めることが可能である。また、f(X₁,X₂)は、細かい有限要素メッシュを用いた場合と粗い有限要素メッシュを用いた場合の、二つの制約条件に関する解析結果の差の関数である。

Here, X ₁ and X ₂ represent the dimension A and the angle A, respectively. Further, a to f mean polynomial coefficients, and these coefficients can be obtained by using the least square method. Further, f (X ₁ , X ₂ ) is a function of the difference between the analysis results regarding the two constraints when the fine finite element mesh is used and when the coarse finite element mesh is used.

In this embodiment, a correction function such as Equation 1 is created by the correction function creation unit 104 and registered in the storage device 109 so that the correction function can be displayed on the display device in response to an operator's request. FIG. 5 shows an example of a correction function display screen. Various methods can be considered to display the correction function. In the example shown in the figure, a preliminary value is obtained when one value arbitrarily selected for the angle A is fixed, the dimension A is the horizontal axis, and the maximum stress is the vertical axis. The fine analysis, that is, the result of numerical analysis by fine finite element mesh and the result of preliminary rough analysis, that is, numerical analysis by coarse finite element mesh, are shown in the upper part of the figure, and the result of preliminary fine analysis and preliminary rough analysis are shown. The correction function obtained as the difference between the analysis results is shown in the lower part of the figure. As seen in the upper part of the figure, there is a large difference (the hatched part in the figure) in the maximum stress obtained by numerical analysis between the fine finite element mesh and the coarse finite element mesh. The difference between the analysis result of the coarse finite element mesh and the analysis result of the fine finite element mesh is obtained in advance as a correction function in the lower part of the figure, and the correction function described below can be performed using the correction function. makes it possible to obtain a high analysis results accuracy performed by the coarse finite element meshes requires only a short processing time of this analysis to derive design solution. This is Phase 2.

The final phase 3 is a phase in which the main analysis for deriving the design solution, the verification analysis for verifying the result, and the like include steps S6 to S10. Step S6 is a main analysis step for executing the optimum calculation based on the optimization condition defined in step S2. The optimization problem exemplified in this embodiment is
Minimization: Mass Condition: 10 mm ≦ X ₁ ≦ 16 mm
0.01 degree ≦ X ₂ ≦ 8 degree
−5 × 9.8 MPa ≦ σ + f (X ₁ , X ₂ ) ≦ 5 × 9.8 MPa
It becomes. Here, sigma represents the stress values obtained by the optimal calculations. Such optimization problems can be solved by using generally known optimization methods such as mathematical methods such as penalty methods and exploratory methods such as simulated annealing methods. In the optimal calculation, the calculation of obtaining the stress value individually while sequentially changing the design variable value within the allowable region is repeatedly performed. The number of repetitions of the calculation is a third number of repetitions larger than the first number of repetitions and the second number of repetitions in the preliminary fine analysis and preliminary rough analysis. In the present invention, for such an optimal calculation, a coarse finite element mesh having the same second division density as the finite element mesh used in the preliminary coarse analysis is used. The corrected stress value sigma as constraints obtained in each analysis to be repeated as the correction function f (X _1, X ₂₎ by _{σ + f (X 1, X} 2), the corrected σ + f (X _1, X The value ₂ ) is the actual constraint value in each analysis.

  Here, as described above, the correction function selects the design variable value at an appropriate interval from the allowable range of the design variable at an appropriate interval and determines the number of iterations (the first number determined by the selected number of the design variable value). Obtained by approximating the polynomial of the difference between the results of the preliminary fine analysis and preliminary coarse analysis by repeating the analysis at the number of iterations of (2) or the second number of iterations) using the method of least squares. Yes, it covers the entire allowable range of design variables, that is, all the design variable values used in the optimal calculation. Therefore, by correcting with such a correction function, the result of the optimum calculation using the coarse finite element mesh can be made to have the same accuracy as the case using the fine finite element mesh. That is, by using a coarse finite element mesh, the processing time can be significantly shortened, and the analysis result can be made highly reliable by correction with a correction function.

In the case of the present embodiment, in the calculation of the correction function of the phase 2, since ε is 20% and the design variables are two of the dimension A and the angle A, X is obtained with 25 design variable values. ^MAX = Max | X ⁱ −X | is satisfied. That is, the preliminary fine analysis and the preliminary coarse analysis fine may be performed 25 times each. The calculation time in the case of the preliminary fine analysis using the fine finite element mesh with the first division density is generally about 1 hour per time. If it is 1 hour per time, the calculation time is 25 in total. It will be time. On the other hand, the preliminary coarse analysis calculation using the coarse finite element mesh with the second division density can be 10 times or more faster than the fine finite element mesh. That is, the second division density is in a relationship of 1/10 or less with respect to the processing time with respect to the first division density. Therefore, if it is 6 minutes per time, the total calculation time for the preliminary coarse analysis will be 2 hours 30 minutes. Therefore, the preliminary analysis for creating the correction function requires a total of 27 hours 30 minutes. On the other hand, the optimal calculations in Phase 3, the repetition number of the iterations is several hundred times from several tens of times is usual. Here, assuming that the number of iterations is 50, a coarse finite element mesh is used for the optimum calculation in the present invention, so the calculation time for the optimum calculation is 5 hours. Therefore, the time required from the creation of the correction function to the end of the optimum calculation is 32 hours and 30 minutes in total, and even if the time required for verification analysis described later is added to this, it is 33 hours and 30 minutes. On the other hand, in the conventional method, although there is no processing for deriving the correction function in the present invention, it takes 50 hours to perform the optimal calculation with a fine finite element mesh. From this, according to the present invention, it can be understood that even when the number of repetitions of the calculation in the optimum calculation is 50, which is relatively small, it is possible to obtain a processing time shortening effect of 16 hours and 30 minutes, which is close to 30%. It can also be seen that the effect of shortening the processing time becomes higher as the number of iterations in the optimum calculation increases.

  The above evaluation is for the case where the repeated analysis in the preliminary fine analysis and the preliminary coarse analysis is performed sequentially. In the present invention, as described above, it is possible to simultaneously perform the repeated analysis in each of the preliminary fine analysis and the preliminary coarse analysis fine. In such a case, the above example may be used. For example, a correction function can be created with an analysis time of 1 hour and 6 minutes, and the time required from the creation of the correction function to the completion of the optimal calculation and its verification can be completed in a total of 7 hours and 6 minutes. The processing time shortening effect can be enhanced.

In step S7, it is determined whether the shape in the final result (design solution) obtained by the optimal calculation in step S6 actually satisfies the constraint conditions. Specifically, the shape obtained as the final result in step S6 is subjected to verification analysis using a fine finite element mesh with a third division density (this may be the same as the first division density). The constraint condition value (stress value in the specific example) obtained by the verification analysis is the constraint condition (−5 × 9.8 MPa ≦ σ + f (X ₁ , X ₂ ) ≦ 5 × 9.8 MPa in the specific example) Determine whether or not If the constraint condition is satisfied in this determination, the result of the optimal calculation transition is displayed as it is in step S10 and the process is terminated. If not satisfied, the transition of the optimal calculation is displayed in step S9 and the correction is made. In order to increase the accuracy of the function, a message prompting the change of the optimization condition such as decreasing ε or increasing the allowable value is displayed. FIG. 6 shows an example of a message display screen. In the example of the figure, the vertical axis represents the objective function and the constraint condition, and the horizontal axis represents the number of iterations of the optimal calculation. Here, the solid line represents the objective function obtained by the iterative calculation, and the broken line represents the constraint condition value obtained by the iterative calculation. In addition, a broken line parallel to the horizontal axis in the drawing represents a boundary line of the constraint condition value, and when the constraint condition value is above the parallel broken line along the horizontal axis, it indicates that the constraint condition is not satisfied. In addition, a cross indicates a numerical analysis result of the constraint condition obtained by the verification analysis. As can be seen from the figure, the calculation using a coarse finite element mesh satisfies the constraint condition in the final solution, but the high-precision verification analysis does not satisfy the constraint condition. In such a case, countermeasures such as increasing the accuracy of the correction function by reducing ε are taken. In step S10, a similar graph is displayed, but no text prompting a warning is displayed.

  In the above embodiment, the case where there are two design variables of the dimension A and the angle A has been described as an example. However, it goes without saying that the present invention can be applied even when the number of design variables is two or more. If there are many types, the number of independent variables in the polynomial of the correction function in the above example can be increased accordingly.

  In the present invention, the correction function is obtained by preliminary analysis with a reduced number of iterations, and the analysis for deriving the design solution is performed with a coarse finite element mesh, and the result of the analysis is corrected with the correction function. I have to. For this reason, it is possible to achieve both a significant reduction in processing time and high analysis accuracy. The present invention can be applied to a design technology field that receives support from digital engineering technology such as CAE, and can greatly contribute to an improvement in efficiency in the design technology field.

It is a figure which shows the whole structure of the shape optimization processing apparatus by one Embodiment with the form of a block diagram. Is a diagram illustrating the flow of shape optimization process performed by the shape optimization processing apparatus of FIG. It is a figure which shows the example of the machine component used as shape optimization object. It is a figure which shows an example of the input screen of optimization conditions. It is a figure which shows an example of the display screen of a correction function. It is a figure which shows an example of a change message display screen.

Explanation of symbols

DESCRIPTION OF SYMBOLS 101 Analysis model definition part 102 Optimization condition definition part 103 Analysis part 104 Correction function creation part 105 Restriction condition value correction part 106 Optimal calculation control part 107 Correction information output part 108 Optimization condition change display part 109 Database 110 Computer

Claims (4)

A shape optimization processing apparatus that performs this analysis after performing a preliminary analysis,
An analysis model definition section for setting an analysis model related to the shape of the design target product;
An optimization condition definition section for setting optimization conditions for the analysis model;
A finite element mesh is created from the analysis model, a preliminary fine analysis is performed by repeating the analysis with the finite element mesh at a first division density at a first iteration number, and a second coarser than the first division density. An analysis unit for performing a preliminary rough analysis in which the analysis at the finite element mesh with the division density is repeated at the first iteration number or a second iteration number different from the first iteration number;
Based on the difference between the result of the preliminary fine analysis and the result of the preliminary rough analysis, a correction function creating unit that creates a correction function;
A constraint value correction unit that corrects each result of the analysis in the finite element mesh with the second division density with the correction function and sets the constraint condition value;
In this analysis, the constraint condition value after correcting the first iteration number and the third iteration number larger than the second iteration number in the preliminary analysis by the constraint value correction unit satisfies the optimization condition. A shape optimization processing apparatus comprising: an optimal calculation control unit that determines whether or not the The shape optimization processing apparatus according to claim 1,
The analysis unit performs verification analysis to analyze with a finite element mesh by a third division density finer than the second division density,
The shape optimization processing device, wherein the optimization calculation control unit determines whether or not a result of the verification analysis satisfies the optimization condition. The shape optimization processing apparatus according to claim 1 or 2,
A shape optimization processing apparatus, comprising: an optimization condition change display unit that displays on the screen when the optimization calculation control unit determines that the optimization condition is not satisfied. The shape optimization processing device according to any one of claims 1 to 3,
The shape optimization processing apparatus, wherein the analysis unit performs the preliminary fine analysis and the preliminary rough analysis by simultaneous processing.

Images