JP4339808B2 - Structure design method - Google Patents

Description

  The present invention sample relates to a method for designing a structure such as a tire, and more particularly to a method for designing a structure that performs optimal design for a plurality of performances using numerical simulation of the structure.

When designing a structure in which steel cords such as tires and rubber members are laminated, an optimum design is performed by simulating a plurality of performances and determining values of design variables.
Even in the case of a tire having a trade-off relationship with performance, the optimum design is performed in consideration of the balance of the performance of a plurality of tires. For example, the performance required for tires includes, for example, steering stability, riding comfort, rolling resistance characteristics, braking characteristics, etc. These performances are controlled by changing the value of the design variable in one direction. Is improved, but has a trade-off relationship such as reduced ride comfort. Thus, when there is a trade-off relationship with performance, the optimum solution for performance exists as a plurality of combinations instead of one.

By the way, in Patent Documents 1 and 2 below, a method for optimally designing a tire using one performance as an objective function is proposed. However, as described above, in order to optimize various performances having a trade-off relationship. By combining a plurality of performances into a single objective function, an optimal solution can be found.
However, this method cannot provide an overall picture of the set of optimal solutions in performance that is in a trade-off relationship, and is provided with this because there is no degree of freedom because the optimal solution is determined pinpoint. The optimal design proposal is a pinpoint one with no degree of freedom.

JP-A-7-164815 Japanese Patent Laid-Open No. 9-323367

  Therefore, the present invention provides a method for designing a structure that can be optimally designed by a designer with a degree of design freedom even in the case of a plurality of performances that have a trade-off relationship in designing a structure including a tire. The purpose is to provide.

  The present invention is a method for designing a structure, wherein at least one of parameters defining the shape, internal structure, or material characteristics of the structure is defined as a design variable, and among physical quantities acting on the structure, A step of determining at least two or more as objective functions, a simulation calculation using the first model or the first simulation approximation formula that defines the value of the design variable as an input, and a second result using the calculation result as an input Calculating a Pareto optimal solution for the objective function by performing a simulation operation using a model or a second simulation approximate expression, and determining the value of each objective function of the Pareto optimal solution as a vector component value, Generating a self-organizing map with information of Pareto optimal solutions Provides a method of designing a structure characterized by having the steps of: generating a self-organizing map imparted with information of the design variables.

At this time, the spatial scale of the structure represented by the first model or the first simulation approximate expression is larger than the spatial scale of the structure represented by the second model or the second simulation approximate expression. Is preferred.
The first simulation approximate expression or the second simulation approximate expression may be an approximate expression created using experimental results. For example, an approximate expression that holds between a design variable and an objective function can be given from an experimental result obtained based on an experimental design.

The structure includes at least a tire, for example, and a tire-equipped vehicle is exemplified.
The structure includes at least a tire, and the step of obtaining the Pareto optimal solution using numerical simulation is performed using a first model that defines the value of the design variable as an input, and the calculation result The first model is a compound model that models the dispersion of the filler reinforcing phase of the tire tread member, and the second model is at least a tire model. A tire model that reproduces a part is preferable.

  Furthermore, in the step of obtaining the Pareto optimal solution, it is preferable to obtain the Pareto optimal solution using a multi-objective genetic algorithm.

In the present invention, using a numerical simulation of a structure such as a tire, a self-organizing map to which information on the Pareto optimal solution for the objective function is added is generated, and a self-organizing map to which information on design variables is added is generated. . For this reason, the designer determines the optimal design plan by determining the position on the self-organizing map while considering the performance balance in a trade-off relationship while looking at the whole picture of the Pareto optimal solution on the surface of these maps. Can be determined. In other words, the optimum design plan can be determined while having a degree of freedom in performance balance.
In particular, a structure represented by the first model or the first simulation approximate expression using the first model or the first simulation approximate expression and the second model or the second simulation approximate expression in the numerical simulation. Because the multiscale simulation in which the spatial scale of the structure is different from the spatial scale of the structure represented by the second model or the second simulation approximation formula, the scale is changed from the characteristic in the microstructure of the structure to the performance of the structure. A consistent analysis can be performed, and an optimal design plan with high accuracy can be determined.

  Hereinafter, a method for designing a structure according to the present invention will be described in detail based on a preferred embodiment shown in the accompanying drawings.

FIG. 1 is a block diagram of a processing apparatus 10 that implements an optimum tire designing method that is an example of a structure designing method according to the present invention.
The processing device 10 includes an input operation system 12, a computer 14, and an output device 16. The input operation system 12 is a mouse or a keyboard, and is a device that inputs various types of information according to instructions from the operator.

The computer 14 includes a CPU 18 and a memory 20, and also includes a ROM (not shown). The computer 14 executes computer software stored in a ROM or the like, thereby executing a condition setting unit 22, an optimum design processing control unit 24, a microscale model creation unit 26, a mesoscale model creation unit 28, and a macroscale model creation unit 30. In this device, each part of the simulation calculation unit 32 and the physical property value adjustment unit 34 is functionally formed, and the self-organization map in which the optimum design plan of the tire is mapped is output to the output device 16. The self-organizing map will be described later.
The output device 16 is a display or printer that displays an input screen so that an operator can instruct using the input operation system 12, and displays various finite element models and simulation calculation results described later.

  The condition setting unit 22 is a part for setting at least two or more objective functions, design variables, and further constraint conditions for performing an optimal design, which are input using the input operation system 12. The objective function is a physical quantity that is to be evaluated as tire performance. For example, a lateral force CP (cornering power) at a slip angle of 1 degree, which is an index of steering stability, and a tire primary index, which is an index of riding comfort. Examples thereof include natural frequency, rolling resistance as an index of rolling resistance, and wear energy of a tire tread member as an index of wear resistance.

The objective function has a preferable direction in terms of performance, and the value increases, decreases, or approaches a predetermined value. Such information on the preferred direction of the objective function is set together with the setting of the objective function. When it is preferable that the objective function approaches a predetermined value, the absolute value of the difference between the objective function value and the predetermined value may be set to be small.
Design variables are parameters that define tire shape, internal structure, and material characteristics. For example, filler dispersion shape and filler volume ratio that define material characteristics in the tread member, and radius of curvature that defines crown shape in the tread portion of the tire. Moreover, the belt width dimension of the tire etc. which prescribe | regulate a tire internal structure are mentioned.
The constraint condition is a condition for constraining the value of the objective function to a predetermined range or constraining the value of the design variable to a predetermined range.
In addition to this, travel conditions such as tire load load and tire rolling speed, road surface conditions (concave / convex shape, coefficient of friction) on which the tire travels, vehicle specification information used for vehicle travel simulation, etc. are set. Is done.

  The optimum design processing control unit 24 sets a sample set in which the values of the design variables are changed variously, inputs the values of the design variables for each sample, performs a simulation calculation described later, and performs at least two objective functions. By searching for a Pareto optimal solution based on multi-objective GA (genetic algorithm), and using this search result, a self-organizing map for the Pareto optimal solution and a self-organizing map with design variable information added This is the part that generates and supplies to the output device 16.

A Pareto optimal solution is a solution in which a plurality of objective functions in a trade-off relationship do not have an advantage over any other solution, but no better solution exists. In general, there are a plurality of Pareto optimal solutions as a set.
The optimum design processing control unit 24 uses a multi-objective GA method for obtaining a set of a plurality of Pareto optimum solutions in one search. A normal GA performs evaluation, selection, crossover, and mutation processing on a sample set of initial solutions, generates a solution set of the next generation, and continues this processing sequentially, after several tens to several hundred generations This is a method for obtaining an optimal solution by obtaining a set of solutions.

  The multi-objective GA is a method for obtaining a boundary of the optimal solution as a Pareto optimal solution by using the GA method because there are a plurality of optimal solutions for a plurality of objective functions. Various methods are currently proposed for the multipurpose GA, but are not particularly limited in the present invention. For example, DRMOGA (Divided Range Multi-Objective GA), NCGA (Neighborhood Cultivation GA), or DCMOGA (Distributed Cooperation model) that divides the solution set into a plurality of regions along the objective function and performs multi-objective GA for each divided solution set. of MOGA and SOGA), NSGA (Non-dominated Sorting GA), NSGA2 (Non-dominated Sorting GA-II), and SPEA II (Strength Pareto Evolutionary Algorithm-II) method can be used. At that time, it is necessary to obtain a set of Pareto optimal solutions with high accuracy, with the solution set widely distributed in the solution space. For this reason, the optimum design processing control unit 24 performs selection using, for example, a vector evaluation genetic algorithm (VEGA), a Pareto ranking method, or a tournament method.

The optimal design processing control unit 24 controls such an optimal design processing routine by the multi-purpose GA, sets a sample set in which each set design variable is variously changed, performs a simulation operation on the sample set, and sets the objective function. By obtaining a value, a sample set of an initial solution of the Pareto optimal solution is set, and multipurpose GA is performed using this set.
Each evaluation, selection, and crossover operation on the solution set in the multi-objective GA is performed according to the value of an objective function that is a simulation calculation result to be described later, and a mutation operation is further performed to generate a next-generation solution set.

As shown in Fig. 2, the self-organizing map is expressed by clustering multi-dimensional input data without any prior knowledge (unsupervised) and divided into multiple regions. It is a map. The value of the objective function in the Pareto optimal solution is used as this multidimensional data.
In the self-organizing map, there are nodes (hexagonal cells in FIG. 2) regularly arranged on a two-dimensional plane, and each of these nodes is a vector of the same number of dimensions as the set number of objective functions. The vector component has a reference vector initialized with a random number. On the other hand, the Pareto optimal solution has a value for each objective function, and when this value is represented by a vector having a vector component, the node corresponding to the reference vector closest to this vector is selected as the best match node (winner). The Then, the vector of the Pareto optimal solution so that the vector of the best match node and the vector of the nodes existing within a predetermined range from the best match node approach the vector of the Pareto optimal solution according to the distance between the vector of the Pareto optimal solution Updated every time. A map composed of nodes having such reference vectors is a self-organizing map. For the self-organization map, see “Visualization and Data Mining of Pareto Solutions Using Self-Organizing Map” (Shigeru Obayashi and Daisuke Sasaki, 2nd International Conference on Evolutionary Multi-Criterion Optimization, 8-11 ^th April 2003, Portugal). Details are stated.

FIG. 2 shows an example of a self-organizing map in the objective functions f ₁ , f ₂ , and f ₃ . In this self-organizing map, it is divided into four areas A to D, and the area A is an area with excellent performance using the objective function f ₁ as an index. The region B is a region where performance using the objective function f ₂ as an index is excellent. The region C is a region where performance using the objective function f ₃ as an index is excellent. Region D is a region in which both performances are inferior. Although not explicitly shown in FIG. 2, each region is displayed in a different color, and the higher the objective function value, the darker the color density.
Further, the optimum design processing control unit 24 generates a self-organizing map in which information on values in a plurality of objective functions is given by color and density, and gives information on values of design variables constituting each Pareto optimum solution. Generated self-organizing map.
Such a self-organizing map is supplied to the output device 16.

On the other hand, the microscale model creation unit 26, the mesoscale model creation unit 28, and the macroscale model creation unit 30 are portions for creating a simulation model for performing a simulation operation in order to calculate the value of the objective function. These simulation models are modeled according to the values of design variables set by the multipurpose GA, and are used for multiscale simulation calculations.
The microscale model creation unit 26 is a part that creates a microscale model that reproduces the microstructure of a rubber member in which fillers are dispersedly arranged. Specifically, when a representative region of a heterogeneous material such as a tire tread rubber member is input from a data supply device (not shown), the microscale model is divided into elements in three directions at regular intervals in each representative region. This is a model in which material constants are assigned to finite elements.

In addition to this, the microscale model may be a microscale model that reproduces a member in which a steel wire is embedded in a rubber member.
The representative region is a region that is set so as to form a lump of heterogeneous material by continuously arranging adjacent regions in three directions orthogonal to each other, and has, for example, a rectangular parallelepiped shape. The created finite element model is composed of a rectangular parallelepiped finite element model. By creating the finite element model, a computable whole matrix representing the finite element model is automatically generated.

  The mesoscale model creation unit 28 is a part that creates a mesoscale model in which a region larger than the representative region represented by the microscale model is reproduced with a finite element. For example, it is a model that reproduces a part of a tread member model of a tire that reproduces a state in which the tread rubber member is deformed in accordance with the minute unevenness when the tread rubber member comes into contact with the minute uneven surface.

The macro-scale model creation unit 30 is a tire overall model in which each tire constituent member is configured by a finite element in order to perform a simulation calculation, and a material that reproduces and gives a tire with a uniform structure of each tire constituent member This is a part for creating a model in which the constant is an equivalent material constant of the tire constituent member.
The microscale model, mesoscale model, and macroscale model will be described later.

The simulation calculation unit 32 is a part that performs a simulation calculation under a predetermined condition using each created simulation model. For example, in a microscale model, a predetermined external force or forced displacement is applied to determine the displacement of each node, a predetermined stress is applied to determine strain, and an external force or stress is applied to determine strain energy. From this, the physical property value (Young's modulus, shear stiffness, Poisson's ratio, or superelastic potential parameter) of the rubber member or the like reproduced by the microscale model is obtained. Alternatively, an external force, displacement, stress, or strain is applied to the microscale model, and a stress / strain analysis simulation is performed using the microscale model to obtain a strain distribution or stress distribution that acts on the microscale model.
The simulation calculation is performed using a known finite element program such as ABQUS.

  In the case of simulation using a mesoscale model, for example, a contact simulation between a tire tread member model and an uneven road surface model under a predetermined condition is performed to obtain a friction coefficient of the tire tread member with respect to the road surface. A tire tread member having a high hardness has a smaller contact area with respect to fine irregularities on the road surface, and the limit friction coefficient is lowered. However, a contact simulation is performed to reproduce such friction characteristics. Also, simulation is performed by giving external force, displacement, stress or strain to the mesoscale model as boundary conditions, contact area in contact with the road surface in the mesoscale model, contact shape, contact pressure distribution, strain energy, stress distribution, strain distribution, etc. Ask for.

  In the case of a simulation using a macroscale model, for example, under a predetermined load condition, the entire tire model which is a macroscale model is brought into contact with a flat road surface model, and further obtained by the above simulation (mesoscale simulation). The tire rolling simulation is performed based on the friction coefficient and the physical property value obtained by the simulation (microscale simulation) using the rubber microstructure model. By this simulation, for example, forces and moments acting on the tire rotation axis of the entire tire model are obtained.

  The physical property value adjustment unit 34 predicts the physical property value used for each model by the simulation operation unit 32 from information such as stress distribution, strain distribution, and strain energy distribution obtained by the simulation operation, and predicts the result. This is the part that adjusts the physical property values used in each model. The prediction of the change in the physical property value is obtained by the simulation calculation unit 32 by examining the correspondence relationship between the degree of wear and the physical property value, the stress distribution, the strain distribution, and the strain energy distribution in advance through experiments and storing them as a database. Predict changes in physical properties from information on stress distribution, strain distribution, and strain energy distribution. The physical property value adjusted in this way is further used as a material constant in a simulation calculation performed using a model.

A tire optimum design method implemented by the processing apparatus 10 will be specifically described.
FIG. 3 is a flowchart showing a flow of an example of an optimum tire design method.

First, on the basis of an instruction input from the input operation system 12, at least two or more objective functions, design variables, and further constraint conditions for optimal design are set (step S10). The design variables are selected from parameters that define the tire shape, internal structure and material properties.
Next, a combination sample set of design variables in which the values of the set design variables are variously determined is determined, and a simulation calculation is performed using this sample set, and a sample set of the initial solution is obtained by obtaining the value of the objective function Ask. With the initial solution sample set as a start, a Pareto optimal solution is searched (step S20).
The search for the Pareto optimal solution is performed by the following multi-scale simulation calculation using the multi-purpose GA by the known method described above.

Multi-scale simulation calculation is performed as follows.
As shown in FIG. 4, a microscale that reproduces a representative region of a rubber member in which granular filler materials such as carbon black and silica and various elastomers are inhomogeneously blended, such as a tire tread rubber member that is a tire constituent member. Create a model, a mesoscale model that models a partial region of a tire that has a larger area than this representative region, and a macroscale model that models the entire tire, and use the simulation calculation results of each scale model as input Perform simulation calculations with models of different scales. These models are all finite element models modeled by a plurality of finite elements.

FIG. 5 is a flowchart showing the flow of multiscale simulation calculation performed to search for the Pareto optimal solution.
As shown in FIG. 4, the tire cross-section 43 of the tire 42 to be mounted on the vehicle 40, the tire tread rubber 45 in which the tire tread rubber area of the tire cross-section 43 is enlarged, and the partial area of the tire tread rubber are enlarged to fill the filler. Each of the rubber microstructures 47 reproducing the microstructure of the rubber material in which the reinforcing phase and the polymer phase are dispersedly arranged is the object of the simulation model. That is, a model of an object corresponding to each scale is created and a simulation is performed. A simulation result is applied to a model of a different scale as a simulation result, and a simulation of a different scale is performed. When the dispersion shape of the filler reinforcement phase, the volume fraction of the filler reinforcement phase, and the radius of curvature of the tire tread are set as design variables for realizing the optimum design, these design variables can be changed to simulation models with different scales. A model is created reflecting the changed value, and each simulation calculation is performed.

First, a rubber microstructure model 46, a tire tread rubber model 48, and an entire tire model 50 of a tire constituent member to be analyzed are created according to design variable values (steps S110, 112, and 114).
A plurality of material phases are dispersedly arranged in the representative region of the tire constituent member to be analyzed, and a material constant is assigned to each finite element of the corresponding rubber microstructure model 50 based on the physical property values of these material phases. Is granted. When the dispersion shape of the filler reinforcing phase and the volume ratio of the filler reinforcing phase are set as the design variables for performing the optimum design, the dispersion shape and volume ratio set as the design variables are used in the model.

FIG. 6A is a view showing a cross section of an example of the rubber microstructure model 50.
FIG. 6A is an example of a cross-sectional view of a rubber microstructure model 50 that models a representative region of a heterogeneous material. In this model, the first polymer phase, the second polymer phase, the granular filler phase (filler reinforcing phase), and the filler-polymer boundary phase surrounding the granular filler phase are inhomogeneous as the material phase. This is a model of representative regions that are distributed. That is, a plurality of elastomers having different material properties are dispersedly arranged as a material phase, and further, granular reinforcing materials (filler reinforcing phases) are dispersedly arranged as a material phase. This heterogeneous material has a maximum elastic modulus of 100 times or more the minimum elastic modulus among the elastic moduli in a plurality of material phases. The material phase constituting the heterogeneous material is not only a solid phase such as an elastomer or a filler, but may also be a void phase filled with gas or liquid.
In the rubber microstructure model 50, a plurality of voxels (unit cells) having the same rectangular parallelepiped (hexahedron) shape whose shape is characterized by a plurality of discrete points are arranged adjacently along three orthogonal directions and continuously. The representative area is reproduced. The representative region is, for example, a rectangular parallelepiped shape, and is divided into meshes in three directions orthogonal to each other, thereby continuously arranging finite elements having a rectangular parallelepiped shape. For example, a finite element model configured by 256 × 256 × 256 finite elements Create Such a finite element model is described in, for example, JP-A-9-180002.

FIG. 6B is a view showing a cross section of an example of the tire tread rubber model 48.
The tire tread rubber model 48 in FIG. 6B has a concave shape by simulating a tire groove, and a model in a state before contact with the uneven road surface model 49 (contact target model) is created. The tire tread rubber model 48 is configured by a finite element model having a rectangular parallelepiped shape (rectangular shape in FIG. 6B). The road surface model 49 is an uneven rigid body model and does not deform. In addition to this, the road surface model 49 may be a model in which a water film and an ice / snow layer are formed on the road surface in accordance with the set road surface conditions.

FIG. 6C is a diagram showing a cross section of an example of the overall tire model 46. The entire tire model 46 is a three-dimensional tire model, and shows a cross section when the entire tire model is cut along a plane perpendicular to the tire rotation axis. The whole tire model 46 is mesh-divided along the shape of the tire constituent member, and is modeled by hexahedral elements, tetrahedral elements, shell elements, membrane elements, and the like.
These scale models are three-dimensional models, but may be two-dimensional models.

  In addition to the model of the rubber material of the tire tread rubber member, the microscale model is a model 52 shown in FIG. 6D, which reproduces a steel belt material formed by coating a steel wire with a rubber material. The model 52 may be a model 54 in which a steel belt material twisted structure as shown in FIG. What is necessary is just to model the tire structural member in which at least the representative region of the tire structural member is composed of a plurality of material phases.

Next, the microscale simulation 1 is executed on the created rubber microstructure model 50 (step S116).
In the microscale simulation 1, the behavior of a node (discrete point) of a finite element located at one end of each arrangement direction of the finite element is smoothly changed to the behavior of a node (discrete point) located at the other end. Periodic boundary conditions are determined so as to be connected, and a predetermined external force is applied under the periodic boundary conditions to determine the stress distribution, strain distribution, or nodal displacement of each finite element. The Young's modulus or shear rigidity in the bulk characteristics of the heterogeneous material is required.

Next, the obtained physical property value is given to the corresponding finite element of the tire tread rubber model 48 which is a mesoscale model, and the mesoscale simulation 2 is performed (step S118).
In the mesoscale simulation 2, the friction characteristic when the tire tread rubber member contacts the road surface is obtained by contacting the road surface model 49 according to the road surface condition set by the tire tread rubber model 48. In this case, when the tire tread rubber model 48 is pressed vertically from above the road surface model 49 or the tire tread rubber model 48 is pressed and brought into contact with the road surface model 49 from the inclined direction, the contact area at this time , Contact pressure, strain energy due to contact, sliding speed, coefficient of friction, etc. are required. The coefficient of friction depends on the physical properties of the rubber material itself in contact with the road surface, but is a characteristic that changes according to the unevenness of the road surface, the contact pressure, and the contact area. Therefore, in the mesoscale simulation 2, this characteristic can be accurately reproduced according to the road surface unevenness. The friction coefficient thus obtained is given to the tire overall model 46 as a characteristic value.

Next, the macro scale simulation 3 is performed using the entire tire model 46 (step S120).
In the macro-scale simulation 3, physical property values such as Young's modulus and shear rigidity obtained in the micro-scale simulation 1 are given as material constants to the tire tread rubber model 48, and further, the friction coefficient is given, and under predetermined conditions. The behavior of the tire in the rolling state of the tire is simulated. The predetermined conditions are, for example, conditions such as an internal pressure when applying an internal pressure to a tire, a load applied to a road surface, a tire rolling speed, a tire slip angle, and a tire camber angle. That is, in the macro-scale simulation 3, a rim model that reproduces a rim created separately is mounted on the entire tire model 46, and the inner peripheral surface of the entire tire model 46 that faces a hollow region surrounded by the entire tire model 46 and the rim model. After that, an internal pressure filling process is performed, and then a load is applied by bringing it into contact with a separately created road surface model 47 (see FIG. 6C). Roll. At that time, a slip angle, a tire camber angle, a rotational torque, or the like is given to the entire tire model 46.

Next, from the calculation results obtained by this simulation, three axial forces (up / down, front / rear, left / right axial forces) acting on the tire rotation axis and moments in the three directions are calculated as tire characteristics (step S122). ).
The axial force and moment may be obtained as time-series data as a physical quantity that changes transiently by giving a predetermined condition to the tire, or the axial force and moment when a steady state is reached. In addition, since the magnitude of the axial force and the moment varies depending on various conditions, for example, the applied load, the tire may calculate the axial force and the moment under various conditions centering on the various set conditions.

Next, a vehicle running simulation is performed using the calculated tire characteristics (step S124). For example, when reproducing a state where a slip angle is given to a tire to generate a lateral force and the vehicle corners and turns, a four-wheel model is created as a vehicle model. For example, a vehicle model is created using a mechanism analysis program ADAMS (manufactured by Mechanical Dynamics. Inc.) and a simulation is performed. Alternatively, a two-wheel model of a vehicle that can be expressed by an analytical expression may be used.
In this way, simulation is performed in the order of the scale from the rubber microstructure model 50 reproducing the microstructure of the tire tread rubber member through the tire tread rubber model 48 and the entire tire model 46, and finally vehicle running simulation is performed. Can do.

In addition, after performing the macro simulation 3, the mesoscale simulation 4 can also be performed (step S126).
That is, a load condition and road contact conditions such as a friction coefficient, a slip angle, and a rolling speed are determined from conditions (reproduction conditions) given to the entire tire model 46, and the macro simulation 3 is performed using these conditions. Thereafter, a mesoscale simulation 4 is performed.
In the mesoscale simulation 4, the tire tread rubber model 48 that has been created and used in the mesoscale simulation 2 with the stress and strain in the corresponding region of the tire tread rubber model 48 calculated in the macroscale simulation 3 is known. Stress or strain is given as a boundary condition to analyze the stress or strain in the tire tread rubber model 48.
Further, the microscale simulation 5 is performed with the stress and strain calculated by the mesoscale simulation 4 being known (step S128).
In the microscale simulation 5, the stress and strain in the corresponding region of the rubber microstructure model used in the microscale simulation 1 are known, and the stress or strain is given as a boundary condition to the rubber microstructure model 50, so that the rubber microstructure model 50 is The stress and strain used are analyzed.

Next, the material constant in each material phase of the rubber microstructure model 50 is adjusted from the calculated stress and strain distribution in the rubber microstructure model 50 (step S130).
The adjustment of the material constant is intended to indicate that the physical property value of each material phase dispersed and arranged as the microstructure of the rubber member changes according to the use situation of the tire. The stress calculated by the rubber microstructure model 50, Material constants (Young's modulus, strain stiffness, Poisson's ratio, etc.) are corrected from the strain distribution. For example, the rubber sample is fatigued by applying a predetermined number of strains and stresses to the rubber sample in advance, and a correspondence table is created that defines the amount of material constants to be corrected based on changes in physical property values accompanying the fatigue. Keep it. By using this correspondence table, the correction amount is obtained from the calculated stress and strain distribution, and the material constant is adjusted.

The adjusted material constant is given again to the rubber microstructure model 50 used in the microscale simulation 1, and the microscale simulation 1 is performed again. Thus, the material constant as the tire tread rubber member is determined based on the physical property values calculated in the microscale simulation 1, and this material constant is given to the mesoscale model to perform the mesoscale simulation 2. Further, the macro school simulation 3 is performed again based on the characteristic values calculated in the mesoscale simulation 2 and the physical property values calculated in the microscale simulation 1. In this way, by performing the macro scale simulation 3 of the tire based on the physical property value of the tire constituent member that changes depending on the use state of the tire, the tire characteristics used in the vehicle running simulation can be calculated. It is possible to reproduce a running simulation of a vehicle that changes depending on usage conditions.
In the method shown in FIG. 5, the value of the objective function is calculated by calculation using a simulation model. However, in the present invention, the value of the objective function may be calculated using a simulation approximate expression. For example, a Pareto optimal solution can be obtained from an experimental result obtained based on an experimental design using an approximate expression (simulation approximate expression) between a design variable and an objective function. As this simulation approximate expression, a known nonlinear function obtained by a polynomial, a neural network or the like can be cited. In the multi-scale simulation calculation, the value of the objective function may be calculated using a simulation model and a simulation approximation formula together.

The result of the simulation calculation that reproduces the running simulation is performed for each combination of design variables with various values changed, the value of the physical quantity of the simulation model set as the objective function is calculated, and a set of initial solutions is generated. .
The set of initial solutions is set as the first generation, and the value of the objective function is obtained using the simulation operation, and the multipurpose GA is obtained until the set number of generations or until the value of the objective function converges within a predetermined range. To obtain the Pareto optimal solution.
The above is the search for the Pareto optimal solution in FIG. 3 (step S20).

Next, a self-organizing map using the value of the objective function is generated using the obtained Pareto optimal solution (step S30). In the self-organizing map, when the value of the objective function is a vector component, the Pareto optimal solutions that are closer to each other are placed closer to each other on the map, and the Pareto optimal solutions are clustered. The
In such a self-organizing map, an area is divided by clustering according to the value of the objective function, and the area is divided as a preferable objective function value by this area, that is, the tire performance is mapped by area. In other words, color display is performed for each region, and by changing the color density so that the color density increases as the value of the objective function goes in the preferred direction, a self-organizing map using color information related to tire performance is generated. Is done. In this way, the Pareto optimal solution set is mapped to the self-organizing map.

Next, a self-organizing map to which information on the values of design variables that achieve the Pareto optimal solution is given is generated (step S40). Since the solution space of the Pareto optimal solution is mapped in the self-organizing map, the value of each design variable that achieves the Pareto optimal solution at this mapping position has a one-to-one correspondence with the mapping position. For this reason, information on the value of the design variable can be given on the self-organizing map by color display or color density.
These self-organizing maps are output to the output device 16 of the display and printer.

FIG. 7 is an example for explaining the self-organizing map in an easy-to-understand manner, and an example of the self-organizing map based on the result of single-scale simulation calculation of the tire performance when the tire performance is defined as an objective function. Show. In the example shown in FIG. 7, the tire tread pattern pitch number, main groove width, and lug groove width of a passenger car tire (tire size: 235 / 45R17) are set as design variables, the objective function is CP (N), and the maximum generated lateral force CF _max (N), DRY braking distance (m), hydroplaning generation speed HP (km / hour), and pattern noise PN (dB). The set design variable is a design variable in the macro scale model. On the other hand, CP (N), CF _max (N), DRY braking distance (m), and HP (km / hour) that are objective functions are calculated using a simulation model, and pattern noise PN (dB) is A value is calculated using a simulation approximation formula that reproduces pattern noise, and is indexed.
FIGS. 8A to 8E are self-organizing maps regarding the optimum pallet solution obtained by using the multi-scale simulation calculation results of tire performance, and information on the index of the value of each objective function is displayed in color. It is a self-organizing map. Below each map, a bar indicating the correspondence between the value in each objective function and the color display (color, color density) is shown. For example, in FIG. 8A, CP variables are displayed in color.

FIGS. 9A to 9C show self-organizing maps to which design variable information is added.
The lower side of each map shows the correspondence between the values of the design variables of the Pareto optimal solution corresponding to the respective positions on the map and the color display (color, color density). In FIG. 9A, the number of pitches is indexed, in FIG. 9B, the main groove width is indexed, and in FIG. 9C, the lug groove width is indexed and displayed in color. .
In this way, the designer can output the self-organization map displaying the objective function value in color and the self-organization map displaying the design variable value in color to the display / printer. Can be visually recognized, and the optimum design plan can be determined based on this map in consideration of the performance balance. Each Pareto optimal solution is an optimal solution that maximizes performance having a trade-off relationship.
For this reason, the designer can design optimally with a degree of freedom of design using these self-organizing maps.

FIGS. 10A to 10C show other examples of the self-organizing map. In the example shown in FIGS. 10A to 10C, the dispersion function of the filler reinforcing phase of the tire tread member of the passenger car tire (tire size: 225 / 45R17) and the crown shape of the tire tread member are set as design variables, and the objective function is used. Is an example of a self-organizing map in which CP is an index of steering stability, a longitudinal spring constant is an index of riding comfort, and a friction energy is an index of wear resistance.
In this case, the design variables are the dispersion shape of the filler reinforcing phase for the microscale model and the crown shape for the macroscale model.

In FIG. 10A, the region where the steering stability is improved with CP as an index is displayed in color density. In FIG. 10A, the lower left region is a region with excellent steering stability.
FIG. 10 (b) displays a region in which riding comfort is improved using the vertical spring constant as an index, in color density. In FIG. 10B, the upper left region is a region having excellent riding comfort.
FIG. 10C shows the area where the wear resistance is improved with the frictional energy as an index, as a color density. In FIG. 10C, the upper right region is a region with excellent wear resistance.
In the conventional optimal design method using a single objective function, as shown in FIG. 11, a design plan in which riding comfort and wear resistance are balanced is determined at the position of point E, and control is performed. A design plan specialized for stability is pinpointed at the position of point F, and a design plan that balances the three performances of steering stability, ride comfort and wear resistance is selected at the point G position . However, when a self-organizing map is used as in the design method of the present invention, information on the optimal solution is given on the surface of the self-organizing map. Using this, the designer can determine the optimum design plan with design freedom.

  The structure designing method of the present invention has been described in detail above. However, the present invention is not limited to the above-described embodiments, and various improvements and modifications may be made without departing from the scope of the present invention. Of course.

It is the block diagram which showed functionally the structure of the processing apparatus which implements the tire design method which is an example of the structure design method of this invention. It is a figure which shows an example of the self-organization map produced | generated with the design method of the structure of this invention. It is a flowchart which shows the flow of the design method of the structure of this invention. It is explanatory drawing explaining the flow of the simulation calculation used for the tire design method which is an example of the structure design method of this invention. It is a flowchart which shows the flow of the simulation calculation shown in FIG. (A)-(e) is a figure which shows the various models used for the simulation calculation shown in FIG. It is a figure explaining the design variable set by the tire design method which is an example of the structure design method of this invention. (A)-(e) is a figure which shows an example of the self-organization map produced | generated with the design method of the structure of this invention. (A)-(c) is a figure which shows the other example of the self-organization map produced | generated with the design method of the structure of this invention. (A)-(c) is a figure which shows the other example of the self-organization map produced | generated with the design method of the structure of this invention. It is a figure which shows the other example of the self-organization map produced | generated with the design method of the structure of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Processing apparatus 12 Input operation system 14 Computer 16 Output apparatus 18 CPU
DESCRIPTION OF SYMBOLS 20 Memory 22 Condition setting part 24 Optimal design process control part 26 Micro scale model creation part 28 Meso scale model creation part 30 Macro scale model creation part 32 Simulation operation part 34 Physical property value adjustment part 40 Vehicle 42 Tire 46 Tire whole model 48 Tire tread Rubber model 50 Rubber microstructure model 52, 54 model

Claims (10)

A structure design method for designing a structure in a structure design system constructed using a computer ,
Structure shape, at least one defined as a parameter sac Chi design variables that define the internal structure or material properties parameter, and at least two defined as the physical quantity sac Chi objective function acting on the structure a physical quantity, comprising the steps to set its setting means to the computer,
The first model or the first model that defines the value of the design variable, which reproduces the spatial scale of a part of the structure of the structure, for calculating the value of the objective function set in the setting step. And the spatial scale of the structure of the structure that includes the part of the structure of the structure represented by the first model or the first simulation approximate expression and is larger than the spatial scale of the part. The creation means of the computer creates a second model or a second simulation approximation formula that reproduces
Computing means of said computer, said in order to calculate the value of the objective function, as well as simulation operation using the creation step has been pre-Symbol first model or the first simulation approximate expression created in this calculation result by simulation operation using the second model or the second simulation approximation equation as an input and calculates a value of the objective function, and obtaining the Pareto optimal solution to the objective function,
Map generating means of the computer, by determining the value of each of the objective function of the Pareto optimal solutions as the value of the vector components, and generating a self-organizing map impart information Pareto optimal solutions,
The map generating means, have a, and generating a self-organizing map giving information of the design variables,
The first simulation approximate expression and the second simulation approximate expression are approximate expressions between the design variable and the objective function on the spatial scales represented by them for obtaining the value of the objective function from the design variable. design method for structures, characterized in that it. The structure design method according to claim 1, wherein the second model or the second simulation approximate expression is reproduced with respect to a spatial scale of the entire structure of the structure. The creating step includes the spatial scale of the first model or structure of the structure of the first simulation approximation formula represents, large compared to the spatial scale, the second model or the second simulation The third model or the third simulation approximation formula reproduced for the spatial scale of the structure of the structure, which is included in the spatial scale of the structure of the structure represented by the approximate expression and has a smaller size than the spatial scale. Create
In the step of obtaining the Pareto optimal solution, the calculation result of the simulation calculation using the first model or the first simulation approximate expression is used as an input to calculate the value of the objective function. The structure according to claim 2 , wherein a simulation calculation is performed using the third simulation approximate expression, and the calculation result is input using the second model or the second simulation approximate expression as an input. Design method. In the step of obtaining the Pareto optimal solution, the simulation calculation using the second model or the second simulation approximate expression is performed, and then the third model or the third model is again used using the calculation result. The simulation calculation is performed using the simulation approximate expression, the simulation result is again performed using the first model or the first simulation approximate expression using the calculation result, and the design variable of the design variable is calculated based on the calculation result. The value is adjusted, the adjusted value of the design variable is input, and the simulation calculation using the first model or the first simulation approximate expression is performed again, and the calculation result is used again. , Performing a simulation operation using the third model or the third simulation approximation formula, Using the calculation result, again, the value of the objective function is calculated by performing a simulation operation using the second model or the second simulation approximation formula, and the Pareto optimal solution for the objective function is calculated. The structure design method according to claim 3, wherein the structure is obtained. The method for designing a structure according to any one of claims 1 to 4, wherein the structure includes at least a tire. The structure includes at least a tire;
Determining the Pareto optimal solutions, using numerical simulation, as well as operation using the first model that defines the value of the design variables as input, using the second model as input the result of the calculation A step of calculating
The first model is a compound model that models dispersion of a filler reinforcing phase of a tire tread member, and the second model is a tire model that includes the first model and reproduces at least a part of the tire. The method for designing a structure according to any one of claims 1 to 4 . The structure design method according to claim 6, wherein the second model is a tire model that reproduces the entire tire. Wherein in the step of obtaining the Pareto optimal solutions, the design method of the structure according to any one of claims 1 to 7, obtaining the Pareto optimal solutions using multipurpose genetic algorithm.   The computer is caused to function as the setting unit, the creation unit, the calculation unit, and the map generation unit, and each step of the structure designing method according to any one of claims 1 to 8 is performed on the computer. A program to be executed.   A computer-readable recording medium on which the program according to claim 9 is recorded.

Images