JP7529985B2 - TIRE SHAPE DETERMINATION METHOD, TIRE SHAPE DETERMINATION DEVICE, AND PROGRAM - Google Patents

Description

The present invention relates to a tire shape determination method in which a computer determines the tire shape, a computer-executable program that causes a computer to determine the tire shape, and a tire shape determination device that determines the tire shape.

Traditionally, tire manufacturers design tires by creating a tire shape that indicates the contour of the tire cross section (cross section along the tire radial direction) using multiple dimensional parameters of the tire. The dimensional parameters include measurable parameters such as the maximum tire outer diameter, maximum tire width, radius of curvature of the tire tread portion, width of the tire tread portion, and maximum tire width position. Such tire shapes are created using a tire shape creation system.

Since the tire shape affects tire performance, various tire shapes are created by changing the values of the dimensional parameters in the tire shape creation system in order to make the tire achieve the desired target tire performance values. Furthermore, a tire model (e.g., a finite element model) having these tire shapes is created, and a simulation calculation of the tire performance is performed, thereby making it possible to search for an optimal tire shape.
On the other hand, there are also cases where various tire shapes are created by changing the values of the non-dimensional parameters in various ways without using conventional dimensional parameters. In this case, a tire model (e.g., a finite element model) having the created tire shape is created, and a simulation calculation of tire performance is performed, thereby allowing an optimum tire shape to be searched for.

For example, a technology is known in which a product shape design method is applied to tire shapes, in which a wide design range is defined with few design variables to efficiently find the optimal product shape that optimizes product performance when optimally designing the product shape based on the evaluation values of product performance (Patent Document 1).

JP 2002-15010 A

The above-mentioned product shape design method includes a shape generation process in which a plurality of base shapes of a product shape, such as a tire shape, are set and a plurality of sample product shapes are generated by linearly combining these base shapes, a performance evaluation process in which an evaluation value of the product performance of the sample product shapes generated by this shape generation process is obtained, and a product shape extraction process in which an optimal product shape with an optimal evaluation value is extracted based on the evaluation value of the product performance obtained by this performance evaluation process.

In the technology of the above-mentioned product shape design method, weighting coefficients related to base shapes used when generating a plurality of sample product shapes by linearly combining base shapes are used instead of dimensional parameters used in the above-mentioned conventional tire shape determination system. The weighting coefficients are non-dimensional parameters that determine the tire shape but cannot be measured.
Although the above-mentioned product shape design method can stably create a tire shape, since the values of the non-dimensional parameters are not measurable dimensions, it is difficult for tire designers to obtain easy-to-understand and useful knowledge for improving tire performance from the values of the non-dimensional parameters. Also, even when the tire shape is formed using control points used in free curves such as Bezier curves and non-uniformly advantageous B-spline (NURBS) curves instead of the weighting coefficients related to the above-mentioned base shapes, it is difficult for tire designers to obtain easy-to-understand and useful knowledge for improving tire performance from the position information of the control points.

Meanwhile, the conventional tire shape creation system described above uses measurable dimensional parameters, so tire designers can obtain easy-to-understand and useful knowledge for improving tire performance from the values of the dimensional parameters. However, depending on the combination of dimensional parameter values, the tire shape creation process performed by the tire shape creation system may partially fail, making it impossible to create the tire shape, and tire shapes cannot be created stably. In addition, tire shapes determined by dimensional parameters are subject to limitations on the degree of freedom, making it impossible to create various tire shapes with sufficient degree of freedom.

The present invention aims to provide a tire shape creation method and device that can stably create a tire shape that can achieve the desired tire performance target values and that is determined by the values of measurable dimensional parameters, and a program that causes a computer to execute the creation of this tire shape.

One aspect of the present invention is a tire shape determination method in which a computer determines a tire shape by activating, within a computer, a first tire shape creation system that creates a tire shape using a plurality of tire dimensional parameters that can be measured with respect to a tire shape that indicates a contour line of a tire cross section, and a second tire shape creation system that creates the tire shape using a plurality of non-dimensional parameters related to the tire shape that cannot be measured with respect to the tire shape. The method includes the steps of: (A) creating a tire model having an original tire shape created by the second tire shape creation system using the non-dimensional parameters;
(B) simulating tire performance using the tire model; and
(C) extracting values of the dimensional parameters related to a reproduced tire shape obtained by reproducing the original tire shape by the first tire shape creation system;
(D) repeating steps (A) to (C) to prepare a plurality of sets of values of the calculation results of the simulation of tire performance and values of the dimensional parameters of the reproduced tire shape, and machine-learning a relationship between the calculation results of the simulation of tire performance and the extracted dimensional parameters into a predictive model;
(E) determining an optimal tire shape that realizes a desired tire performance target value by calculating values of the dimensional parameters corresponding to the target value using the machine-learned predictive model; and
has.

Another aspect of the present invention is a tire shape determination method in which a computer determines a tire shape by activating within a computer a first tire shape creation system that creates a tire shape using a plurality of dimensional parameters of the tire that can be actually measured with respect to a tire shape that indicates a contour line of a tire cross section, and a second tire shape creation system that creates the tire shape using a plurality of non-dimensional parameters of the tire shape that cannot be actually measured with respect to the tire shape.
(A) creating an original tire shape using the non-dimension parameters by the second tire shape creation system;
(B) extracting values of the dimensional parameters related to a reproduced tire shape obtained by reproducing the original tire shape by the first tire shape creation system;
(C) creating a tire model having a reproduced tire shape by reproducing the created original tire shape using the dimensional parameters in the first tire shape creation system;
(D) simulating tire performance using the tire model; and
(E) repeating steps (A) to (D) to prepare a plurality of sets of values of the calculation results of the simulation of tire performance and the values of the dimensional parameters related to the reproduced tire shape obtained in step (B), and machine learning the relationship between the calculation results of the simulation of tire performance and the extracted dimensional parameters into a prediction model;
(F) determining an optimal tire shape that realizes a target value of a desired tire performance by calculating values of the dimensional parameters corresponding to the target value using the machine-learned predictive model; and
has.

In extracting the values of the dimensional parameters, it is preferable to use the error between the contour line of the tire shape created by the first tire shape creation system and the contour line of the original tire shape as an objective function, and to extract the values of the dimensional parameters as a minimization problem with the values of the dimensional parameters as unknown variables.

The original tire shape is preferably the shape of the outer surface of the tire.

It is preferable to prepare multiple sets of values of the simulation results of the tire performance and values of the dimensional parameters by adjusting the values of the non-dimensional parameters used in (A) according to the values of the simulation results each time the simulation is repeated so that the values of the simulation results approach the target values.

It is preferable that the simulation calculation of the tire performance includes at least one of the following: calculation of the tire shape that changes when the tire is filled with air, calculation of the contact shape when the tire contacts the ground, calculation of the contact pressure distribution when the tire contacts the ground, calculation of the tire spring constant, and calculation of the rolling resistance.

Another aspect of the present invention is a program that causes a computer to execute the tire shape determination method.

Yet another aspect of the present invention is a tire shape determination device that determines a tire shape using a first tire shape creation system that creates a tire shape using a plurality of dimensional parameters of a tire that can be actually measured with respect to a tire shape that indicates a contour line of a tire cross section, and a second tire shape creation system that creates the tire shape using a plurality of non-dimensional parameters related to the tire shape that cannot be actually measured with respect to the tire shape.
a shape creation unit that creates an original tire shape using the non-dimension parameters by the second tire shape creation system;
a tire model creation unit that creates a tire model having the created original tire shape;
a simulation calculation unit that simulates tire performance using the tire model;
a parameter extraction unit that extracts values of the dimensional parameters related to a reproduced tire shape obtained by reproducing the original tire shape created by the shape creation unit using the first tire shape creation system;
a prediction model creation unit that prepares a plurality of sets of values of the calculation results of the simulation of the tire performance and values of the dimensional parameters by repeatedly performing the creation of the tire model by the tire model creation unit, the simulation of the tire performance by the simulation calculation unit, and the extraction of values of the dimensional parameters by the parameter extraction unit, and machine learning the relationship between the calculation results of the simulation of the tire performance and the extracted dimensional parameters into a prediction model;
a shape determination unit that determines an optimal tire shape for achieving a desired tire performance target value by calculating values of the dimensional parameters using the machine-learned predictive model; and
has.

Yet another aspect of the present invention is a tire shape determination device that determines a tire shape using a first tire shape creation system that creates a tire shape using a plurality of dimensional parameters of a tire that can be actually measured with respect to a tire shape that indicates a contour line of a tire cross section, and a second tire shape creation system that creates the tire shape using a plurality of non-dimensional parameters of the tire shape that cannot be actually measured with respect to the tire shape.
a shape creation unit that creates an original tire shape using the non-dimension parameters by the second tire shape creation system;
a parameter extraction unit that extracts values of the dimensional parameters related to a reproduced tire shape obtained by reproducing the original tire shape created by the shape creation unit using the first tire shape creation system;
a model creation unit that creates a tire model having a reproduced tire shape by reproducing the created original tire shape by using the dimensional parameters in the first tire shape creation system;
a simulation calculation unit that simulates tire performance using the tire model;
a prediction model creation unit that prepares a plurality of sets of values of the calculation results of the simulation of the tire performance and values of the dimensional parameters related to the reproduced tire shape by repeatedly performing creation of the tire shape by the shape creation unit, creation of the tire model by the model creation unit, and simulation of the tire performance by the simulation calculation unit, and machine learning the relationship between the calculation results of the simulation of the tire performance and the dimensional parameters into a prediction model;
a shape determination unit that determines an optimal tire shape for achieving a desired tire performance target value by calculating values of the dimensional parameters using the machine-learned predictive model; and
has.

The tire shape creation method, tire shape creation device, and program described above can stably create a tire shape that can achieve the desired tire performance target values and is determined by the values of measurable dimensional parameters.

1 is a diagram illustrating a tire shape determination method according to an embodiment. 1 is a diagram illustrating an example of dimensional parameters used by a first tire shape creation system used in a tire shape determination method of one embodiment. FIG. FIG. 11 is a diagram illustrating an example of a non-dimensional parameter used by a second tire shape creation system used in a tire shape determination method according to one embodiment. 2 is a diagram illustrating a tire shape determination method according to another embodiment different from the tire shape determination method shown in FIG. 1 . 1 is a configuration diagram of a tire shape determination device according to an embodiment; FIG. 2 is a diagram showing an example of a flow of a tire shape determination method according to an embodiment. FIG. 11 is a diagram showing an example of a flow of a tire shape determination method according to another embodiment.

The tire shape determination method, tire shape determination device, and program of the present invention will be described with reference to the drawings. Figure 1 is a diagram illustrating a tire shape determination method of one embodiment.

(Tire Shape Determination Method)
The tire shape determination method shown in FIG. 1 is a method in which a computer determines a tire shape using a first tire shape generation system and a second tire shape generation system running within a computer.
The first tire shape creation system creates a tire shape using a plurality of tire dimensional parameters that can be actually measured with respect to the tire shape. The second tire shape creation system creates a tire shape using a plurality of non-dimensional parameters that cannot be actually measured with respect to the tire shape. The tire shape is the shape of a contour line of a tire cross section cut with a warp in the tire radial direction.

Fig. 2 is a diagram for explaining an example of dimensional parameters used by the first tire shape creation system. As shown in Fig. 2, the dimensional parameters are dimensions that define the contour line of the outer surface of the tire. The dimensional parameters include at least the maximum tire outer diameter DA, the tire inner diameter DC, the maximum tire width position DB, the maximum tire width WA, the tire tread curvature radii RA, RB, ..., and the tire side curvature radii RC, RD, ....
The values of such parameters can be obtained by measuring the tire shape with a laser displacement meter. The tire shape can be created using such dimensional parameters.

Figure 3 is a diagram illustrating an example of non-dimensional parameters used by the second tire shape creation system. The second tire shape creation system creates a composite tire shape obtained by averaging a plurality of predetermined base tire shapes with a weighting coefficient. In Figure 3, the base tire shape is the tire shape of two base tire models BM1 and BM2 that have different predetermined tire shapes but the same model configuration. The base tire models BM1 and BM2 are, for example, finite element models, and the position coordinates of each node and the nodes (node numbers) that constitute each element are determined. Therefore, a composite tire model CM having a new tire shape can be created by multiplying each of the tire models BM1 and BM2 by a weighting coefficient and averaging them (weighting the values of the position coordinates). In the example shown in FIG. 3, a composite tire model CM1 in which the weighting coefficient P1 for the base tire model BM1 is 0.2 and the weighting coefficient P2 for the base tire model BM2 is 0.8, and a composite tire model CM2 in which the weighting coefficient P1 for the base tire model BM1 is 0.4 and the weighting coefficient P2 for the base tire model BM2 is 0.6 are shown. It is not necessary to create a tire shape using a tire model as shown in FIG. 3. A tire shape can be created freely by preparing a plurality of base tire shapes in advance, freely setting the values of the weighting coefficients for each base tire shape, and averaging each base tire shape. Such weighting coefficients P1 and P2 are non-dimensional parameters related to the tire shape that cannot be measured. In FIG. 3, two base tire shapes are used, but three or more base tire shapes can be used.

In the second tire shape creation system shown in FIG. 3, the weighting coefficients for each of the multiple base tire shapes are used as non-dimensional parameters, but when the tire shape is defined by free curves such as Bezier curves and non-uniformly-advantaged B-spline (NURBS) curves, the position coordinates of the control points on these free curves can also be used as non-dimensional parameters.

Returning to FIG. 1, first, the values of the non-dimensional parameters are determined in the second tire shape system, and an original tire shape is created, for example, by a method such as that shown in FIG. 3, and a tire model having this original tire shape (for example, a tire model created by the finite element method) is created. Using this tire model, a simulation calculation is performed to evaluate tire performance. Examples of this simulation calculation include, for example, a calculation of the deformation of the tire shape when the tire is filled with a predetermined internal pressure, a calculation of the contact shape when the tire is grounded, a calculation of the contact pressure distribution when the tire is grounded, a calculation of the tire spring constant, and a calculation of the rolling resistance. The deformation of the tire shape, the contact shape when the tire is grounded, and the contact pressure distribution when the tire is grounded affect the durability performance, wear performance, uneven wear brain, or the tire's motion performance, and are effective evaluation elements in evaluating tire performance. In the case of a tire shape deformation calculation, for example, the tire deformation amount, strain, or stress at a position of interest is calculated. In the case of a contact shape calculation, for example, a value obtained by quantifying the characteristics of the contact shape such as the aspect ratio and shape (rectangular shape, circular shape), etc. In calculating the contact pressure distribution, for example, the ratio of the contact pressures at two points of interest is calculated. Of course, as a simulation calculation, the tire performance can be calculated by performing a process of rolling the tire on the ground and rolling, and calculating the rolling resistance, frictional force, frictional energy, the force and moment generated when cornering, and the vibration when stepping on a protrusion. This type of simulation calculation is a well-known technique, so a description thereof will be omitted.

On the other hand, in the first tire shape creation system, when creating a reproduced tire that reproduces the original tire shape created by the second tire shape creation system using non-dimensional parameters, values of dimensional parameters related to the reproduced tire shape are extracted. As shown in Fig. 1, the dimensional parameter values include the tire maximum outer diameter DA, tire inner diameter DC, tire maximum width position DB, tire maximum width WA, tire tread radii of curvature RA, RB, ..., tire side radii of curvature RC, RD, .... Therefore, by obtaining the values of the dimensional parameters related to the reproduced tire shape created by the first tire shape creation system, a set of the simulation calculation results and the values of the dimensional parameters can be obtained.
In the second tire shape creation system, a variety of different original tire shapes can be created by changing the values of the non-dimensional parameters in various ways, so that multiple sets can be prepared by combining the simulation calculation results obtained from each of these original tire shapes with the values of the dimensional parameters.

These multiple sets are then given as learning data for machine learning into a predictive model that has been set up in advance in a computer. This allows the predictive model to learn the relationship between the tire performance simulation calculation results and the dimensional parameters by machine learning. Predictive models include models using neural networks, such as the well-known deep learning, models using the well-known random forest method that uses multiple decision trees for "classification" or "regression," and models using LASSO regression. In addition, nonlinear functions using polynomials, kriging, or RBF (Radial Base Function) can also be used as predictive models.

Using the machine learning prediction model, the system determines the optimal tire shape that achieves the target value by calculating the dimensional parameter values corresponding to the target value from the target value of the desired tire performance. Furthermore, the system can design the tire by determining the arrangement of the tire components and the thickness and width of the tire components for the optimal tire shape that has been determined.

When a prediction model is constructed to predict simulation calculation results using dimensional parameter values as input data, even if a target value for tire performance is given to the prediction model, it is not possible to predict and output the value of the dimensional parameter corresponding to this target value. In this case, it is preferable to search for the optimal value of the dimensional parameter that achieves the target value by changing the value of the dimensional parameter in various ways. For example, an evolutionary algorithm can be used to search for such an optimal value of the dimensional parameter. Evolutionary algorithms include Genetic Algorithm, Differential Evolution, Particle Swarm Optimization, Ant Colony Optimization, etc. Experimental design and Latin hypercube method can also be used.

As described above, in the first tire shape creation system, depending on the combination of the values of the dimensional parameters, the tire shape creation process may partially fail, making it impossible to create the tire shape, and the tire shape may not be created stably. However, in the second tire shape creation system, the original tire shape can be stably created using non-dimensional parameters, and the first tire shape creation system creates the most approximate (reproduced) reproduced tire shape from this original tire shape using dimensional parameters, so the values of the dimensional parameters can be stably extracted. Moreover, in the second tire shape creation system, when various tire shapes are created using two or more base tire models having a base tire shape such as that shown in Figure 3, the tire shape can be changed with a high degree of freedom, so a tire shape corresponding to the target value can be reliably obtained.

FIG. 4 is a diagram for explaining a tire shape determining method according to another embodiment different from the tire shape determining method shown in FIG.
In the tire shape determination method shown in FIG. 4 , after an original tire shape is created by the second tire shape creation system, this original tire shape is applied to the first tire shape creation system, and dimensional parameter values for reproducing the applied original tire shape by the first tire shape creation system, such as dimensional parameter values such as the tire maximum outer diameter DA, tire inner diameter DC, and maximum tire width position DB, are extracted to create a reproduced tire shape that reproduces the original tire shape.
A tire model (e.g., a tire model created by the finite element method) having a reproduced tire shape is created by reproducing the original tire shape using a first tire shape creation system, and this tire model is used to perform simulation calculations of tire performance in the same manner as the tire shape determination method shown in Figure 1.

In the second tire shape creation system, a variety of different original tire shapes can be created by changing the values of the non-dimensional parameters in various ways, so multiple sets can be prepared by combining the values of the dimensional parameters related to the reproduced tire shape created by the first tire shape creation system from each of these original tire shapes with the calculation results of a simulation using a tire model.
The multiple sets prepared in this way are used as training data for machine learning of a predictive model.
The subsequent flow is the same as that of the tire shape determination method shown in FIG. 1, and therefore a description thereof will be omitted.

In the tire shape determination method shown in FIG. 4, a set of dimensional parameter values of the reproduced tire shape created by the first tire shape creation system and simulation calculation results using a tire model having this reproduced tire shape is used as learning data for the prediction model so as to reproduce the original tire shape created by the second tire shape creation system. Therefore, the prediction accuracy of the prediction model can be improved compared to the case in which the tire shape determination method shown in FIG. 1 uses a set of dimensional parameter values of the reproduced tire shape created by the first tire shape creation system and simulation calculation results using a tire model having the original tire shape as learning data for the prediction model.

(Tire Shape Determining Device)
Fig. 5 is a configuration diagram of a tire shape determining device of one embodiment that performs the tire shape determining method shown in Figs. 1 and 4. The tire shape determining device 10 is configured by a computer including a CPU 12, a RAM 14, and a ROM 16. A mouse/keyboard 17 and a display 18 are connected to the tire shape determining device 10.
The tire shape determination device 10 reads out and starts a program stored in the ROM 16, thereby forming a module 20. In other words, the module 20 is a software module. The RAM 14 temporarily stores intermediate results of the processing described below.

The module 20 includes a shape creation unit 22, a tire model creation unit 24, a simulation calculation unit 26, a parameter extraction unit 28, a predictive model creation unit 30, and a shape determination unit 32. These functions are essentially realized mainly by the calculations of the CPU 12.

The shape creation unit 22 includes a first tire shape creation system and a second tire shape creation system as subprogram modules, and creates a reproduced tire shape using dimensional parameters by the first tire shape creation system in response to instructions from the CPU 12, and creates an original tire shape using non-dimensional parameters by the second tire shape creation system.

The tire model creation unit 24 is a part that creates a tire model so as to have the tire shape created by the shape creation unit 22, i.e., the original tire shape or the reproduced tire shape. The tire model is, for example, a model in which nodes and finite elements are defined. The tire model may be any model that allows numerical calculations, and may be, for example, a finite element model or a discretized model that reproduces a tire using the finite difference method or the finite volume method. Tire models are created for multiple tire shapes, but the types, arrangements, and dimensions of the tire components in the tire model are all fixed so that they are the same.

The simulation calculation unit 26 is a part that performs simulation calculations of tire performance using the created tire model. The simulation calculation is specified by inputting from the mouse/keyboard 17. The simulation calculation includes at least calculation of deformation of the tire shape when the tire is filled with a predetermined internal pressure, calculation of the contact shape when the tire contacts the ground, calculation of the contact pressure distribution when the tire contacts the ground, calculation of the tire spring constant, and calculation of the rolling resistance. Furthermore, as the simulation calculation, a process of rolling the tire on the ground and rolling may be performed to calculate the rolling resistance, frictional force, frictional energy, force and moment generated during cornering, vibration when stepping on a protrusion, etc.
As a result of such simulation calculations, numerical values for evaluating tire performance are calculated.

The parameter extraction unit 28 is a part that extracts values of dimensional parameters related to the reproduced tire shape by reproducing the original tire shape created by the second tire shape creation system using the first tire shape creation system. Here, reproduction of the original tire shape means creating a reproduced tire shape that is closest to or identical to the original tire shape.

The predictive model creation unit 30 prepares a plurality of sets by repeating the process while changing the values of the non-dimensional parameters, using as a set the numerical values of the calculation results of the simulation of a tire model having an original tire shape or a reproduced tire shape created by the shape creation unit 22 based on the values of the non-dimensional parameters and the values of the dimensional parameters related to the reproduced tire shape extracted by the parameter extraction unit 28. The predictive model creation unit 30 assigns the prepared plurality of sets to the predictive model as learning data for machine learning the predictive model. The mouse/keyboard 17 is used to appropriately input instructions as to how to change the values of the non-dimensional parameters in order to create the learning data.
In the case of a prediction model using a neural network, for example, the number of layers and nodes are appropriately set and the model is constructed. The number of layers is, for example, specified and input from the mouse/keyboard 17. The number of nodes changes depending on the construction of the model. In this way, the prediction model becomes a model that has been machine-learned to learn the relationship between the dimensional parameters and the results of the simulation calculations.

The shape determination unit 32 is a part that determines the optimal tire shape that realizes the target value by calculating the value of the dimensional parameters using a machine-learned prediction model from the target value of the desired tire performance input from the mouse/keyboard 17. The optimal tire shape that realizes the target value is determined by the dimensional parameters. Meanwhile, since the prediction model outputs the value of the simulation calculation result by inputting the value of the dimensional parameters, it is necessary to perform a process of searching for the optimal value of the dimensional parameters that achieves this target value from the target value of the tire performance. For this reason, an evolutionary algorithm can be used to calculate the optimal value of the dimensional parameters. The method of searching for such an optimal value is set by inputting instructions from the mouse/keyboard 17.

The display 18 displays the results of processing in the module 20 and information during processing, and can also display an input setting screen for input from the mouse and keyboard 17. The display 18 displays, for example, the original tire shape and the reproduced tire shape, and also displays the tire shape determined by the shape determination unit 32, as well as the values of dimensional parameters.

Figure 6 shows an example of the flow of the tire shape determination method shown in Figure 1.

First, the shape creation unit 22 creates an original tire shape using the values of the non-dimensional parameters by the second tire shape system, and then the tire model creation unit 24 creates a tire model having the created original tire shape (step ST10).

Next, the simulation calculation unit 26 performs a simulation calculation of the set tire performance using the created tire model (step ST12). As a result of the simulation calculation, an evaluation value for the tire performance is calculated.

The parameter extraction unit 28 extracts values of dimensional parameters related to the reproduced tire shape that reproduces the original tire shape created in step ST10 using the first tire shape creation system (step ST14). In the first tire shape creation system, the values of the dimensional parameters are variously changed so that the reproduced tire shape (contour line) optimally approximates the original tire shape (contour line) created in step ST10, and optimal dimensional parameter values are extracted.

The prediction model creation unit 30 prepares the calculation results of the simulation in step ST12 and a set of the dimensional parameter values in step ST14 as one piece of learning data (step ST16).
Next, the prediction model creation unit 30 determines whether or not the number of learning data (number of sets) has reached a predetermined number (step ST18). If the number of learning data has not reached the predetermined number, the values of the non-dimension parameters are adjusted (step ST20), and the process returns to step ST10, where an original tire shape is created by the second tire shape creation system. In this manner, steps ST10 to ST20 are repeated and learning data is prepared until the number of learning data reaches the predetermined number. The predetermined number is, for example, several thousand to several tens of thousands.

When the number of pieces of training data reaches a predetermined number, the prediction model creation unit 30 supplies the training data to the prediction model, and performs machine learning on the prediction model (step ST22).
Thereafter, the machine-learned predictive model is used to search for and determine an optimal tire shape that realizes the target tire performance values inputted through the mouse/keyboard 17 (step S24). When obtaining an optimal tire shape that realizes the target tire performance values, for example, a well-known evolutionary algorithm can be used to find optimal values that are closest to the target values while variously changing the values of dimensional parameters.
The optimum tire shape that realizes the target values is then displayed on the display 18 and sent to a CAD system (not shown), where the types of tire constituent components, the dimensions of the tire constituent components, the positions of the tire constituent components, etc. are considered and the tire is designed.

Figure 7 shows an example of the flow of the tire shape determination method shown in Figure 4.

The contents of steps ST52 to ST54 in the flow shown in FIG. 7 are the same as steps ST22 to ST24 in the flow shown in FIG. 6, so their explanation is omitted.

First, the shape creation unit 22 creates an original tire shape using the values of the non-dimensional parameters with the second tire shape system (step ST40).

Furthermore, the shape creation unit 22 and the parameter extraction unit 28 work in conjunction with each other to extract values of dimensional parameters related to a reproduced tire shape in which the original tire shape created by the second tire shape creation system is reproduced by the first tire shape creation system to create a reproduced tire shape, and further the tire model creation unit 24 creates a tire model having the reproduced tire shape (step ST42). When the reproduced tire shape is created by the first tire shape creation system, the parameter extraction unit 28 extracts values of dimensional parameters. Therefore, when the original tire shape is reproduced by the first tire shape creation system, the values of dimensional parameters related to the reproduced tire shape can also be obtained.
In addition, the original tire shape does not completely match the reproduced tire shape, and a part of the reproduced tire shape may be changed from the original tire shape. For this reason, a tire model having a reproduced tire shape is used in the simulation calculation.

The simulation calculation unit 26 performs a simulation calculation of the set tire performance using the created tire model (step ST44). As a result of the simulation calculation, an evaluation value for the tire performance is calculated.

The prediction model creation section 30 prepares the result of the simulation calculation in step ST44 and a set of the dimensional parameter values obtained in step ST42 as one piece of learning data (step ST46).
In this manner, the prediction model creation unit 30 repeats steps ST40 to ST50 to prepare training data until the number of training data reaches a predetermined number. The predetermined number is, for example, several thousand to several tens of thousands.

The learning data thus obtained is used to train a predictive model, and the machine-learned predictive model can be used to determine the optimal tire shape that achieves the target tire performance values (steps ST52, ST54).

In this way, the values of the dimensional parameters related to the reproduced tire shape that reproduces the original tire shape and the values of the calculation results of a simulation using a tire model having the reproduced tire shape are prepared as learning data, so that a tire shape that can achieve the desired target values of tire performance and is determined by the values of measurable dimensional parameters can be stably created.

According to one embodiment, in the extraction of the dimensional parameter values, it is preferable to use the error between the contour of the tire shape created by the first tire shape creation system and the contour of the original tire shape created by the second tire shape creation system as the objective function, and to extract the dimensional parameter values as a minimum problem with the dimensional parameter values as unknown variables. In this case, to extract the dimensional parameter values, for example, an evolutionary algorithm including a genetic algorithm, differential evolution, particle swarm optimization, or ant colony optimization, or a particle swarm optimization method is used. Experimental design and Latin hypercube methods can also be used. The error between the contours can be the sum of the squares of the distance between the nearest neighboring positions, the sum of the squares of the distance between the contours in a direction perpendicular to the contour of the tire shape created by the first tire shape system, or the sum of the squares of the distance between the contours in the tire radial direction and the squares of the distance in the tire width direction.

The optimum tire shape can be the shape of the inner surface of the tire or the shape of the outer surface of the tire, but according to one embodiment, the optimum tire shape is preferably the shape of the outer surface of the tire. The shape of the outer surface of the tire can provide useful information for the design parameters related to the mold when manufacturing the tire.

According to one embodiment, it is preferable to prepare multiple sets of values of the simulation results of tire performance and values of the dimensional parameters by adjusting the values of the non-dimensional parameters used in creating the original tire shape by the second tire shape system according to the values of the simulation results each time the simulation is repeated so that the values of the simulation results approach predetermined target values. Such processing may involve using a genetic algorithm to set the adjusted values of the non-dimensional parameters. When there are multiple target values of tire performance, a multi-objective genetic algorithm may be used.
In one embodiment, a plurality of sets of learning data may be prepared by creating a plurality of original tire shapes at once using the second tire shape system. In this case, the values of a plurality of non-dimensional parameters defining the original tire shape may be set according to an orthogonal array in an experimental design method, or according to a Monte Carlo method or a Latin hypercube method.

According to one embodiment, the tire performance simulation calculations preferably include at least one of the following: calculation of the tire shape that changes when the tire is filled with air, calculation of the contact shape when the tire contacts the ground, calculation of the contact pressure distribution when the tire contacts the ground, calculation of the tire spring constant, and calculation of the rolling resistance. This makes it possible to establish a relationship between tire performance and dimensional parameters.

Such a tire shape determination method can be achieved by causing a computer to execute an executable program.
Therefore, the program describes the procedure for carrying out the flow shown in FIG. 6 or FIG.

The tire shape determination method, tire shape determination device, and program of the present invention have been described in detail above, but the present invention is not limited to the above-described embodiment, and various improvements and modifications may be made without departing from the spirit of the present invention.

10 Tire shape determination device 12 CPU
14 RAM
16 ROM
17 Mouse/keyboard 18 Display 20 Module 22 Shape creation unit 24 Tire model creation unit 26 Simulation calculation unit 28 Parameter extraction unit 30 Prediction model creation unit 32 Shape determination unit

Claims (9)

A tire shape determination method in which a computer determines a tire shape by activating, within a computer, a first tire shape creation system that creates a tire shape using a plurality of tire dimensional parameters that can be actually measured with respect to a tire shape indicating a contour line of a tire cross section, and a second tire shape creation system that creates the tire shape using a plurality of non-dimensional parameters that cannot be actually measured with respect to the tire shape,
(A) creating a tire model having an original tire shape created by the second tire shape creation system using the non-dimension parameters;
(B) simulating tire performance using the tire model; and
(C) extracting values of the dimensional parameters related to a reproduced tire shape obtained by reproducing the original tire shape by the first tire shape creation system;
(D) repeating steps (A) to (C) to prepare a plurality of sets of values of the calculation results of the simulation of tire performance and values of the dimensional parameters of the reproduced tire shape, and machine-learning a relationship between the calculation results of the simulation of tire performance and the extracted dimensional parameters into a predictive model;
(E) determining an optimal tire shape that realizes a desired tire performance target value by calculating values of the dimensional parameters corresponding to the target value using the machine-learned predictive model; and
A tire shape determination method comprising the steps of: A tire shape determination method in which a computer determines a tire shape by activating, within a computer, a first tire shape creation system that creates a tire shape using a plurality of tire dimensional parameters that can be actually measured with respect to a tire shape indicating a contour line of a tire cross section, and a second tire shape creation system that creates the tire shape using a plurality of non-dimensional parameters that cannot be actually measured with respect to the tire shape,
(A) creating an original tire shape using the non-dimension parameters by the second tire shape creation system;
(B) extracting values of the dimensional parameters related to a reproduced tire shape obtained by reproducing the original tire shape by the first tire shape creation system;
(C) creating a tire model having a reproduced tire shape by reproducing the created original tire shape using the dimensional parameters in the first tire shape creation system;
(D) simulating tire performance using the tire model; and
(E) repeating steps (A) to (D) to prepare a plurality of sets of values of the calculation results of the simulation of tire performance and the values of the dimensional parameters related to the reproduced tire shape obtained in step (B), and machine learning the relationship between the calculation results of the simulation of tire performance and the extracted dimensional parameters into a prediction model;
(F) determining an optimal tire shape that realizes a target value of a desired tire performance by calculating values of the dimensional parameters corresponding to the target value using the machine-learned predictive model; and
A tire shape determination method comprising the steps of: The tire shape determination method according to claim 1 or 2, wherein in extracting the values of the dimensional parameters, the error between the contour of the tire shape created by the first tire shape creation system and the contour of the original tire shape is used as an objective function, and the values of the dimensional parameters are extracted as a minimization problem with the values of the dimensional parameters as unknown variables. The tire shape determination method according to any one of claims 1 to 3, wherein the original tire shape is the shape of the outer surface of the tire. The tire shape determination method according to any one of claims 1 to 4, wherein the values of the non-dimensional parameters used in (A) are adjusted according to the values of the calculation results of the simulation each time the simulation is repeated so that the values of the calculation results of the simulation approach the target values, thereby preparing multiple sets of the values of the calculation results of the simulation of the tire performance and the values of the dimensional parameters. The tire shape determination method according to any one of claims 1 to 5, wherein the calculation of the simulation of the tire performance includes at least one of the following: calculation of the tire shape that changes when the tire is filled with air, calculation of the contact shape when the tire contacts the ground, calculation of the contact pressure distribution when the tire contacts the ground, calculation of the tire spring constant, and calculation of the rolling resistance. A program that causes a computer to execute the tire shape determination method described in any one of claims 1 to 6. A tire shape determination device that determines a tire shape using a first tire shape creation system that creates a tire shape using a plurality of dimensional parameters of a tire that can be actually measured with respect to a tire shape indicating a contour line of a tire cross section, and a second tire shape creation system that creates the tire shape using a plurality of non-dimensional parameters related to the tire shape that cannot be actually measured with respect to the tire shape,
a shape creation unit that creates an original tire shape using the non-dimension parameters by the second tire shape creation system;
a tire model creation unit that creates a tire model having the created original tire shape;
a simulation calculation unit that simulates tire performance using the tire model;
a parameter extraction unit that extracts values of the dimensional parameters related to a reproduced tire shape obtained by reproducing the original tire shape created by the shape creation unit using the first tire shape creation system;
a prediction model creation unit that prepares a plurality of sets of values of the calculation results of the simulation of the tire performance and values of the dimensional parameters by repeatedly performing the creation of the tire model by the tire model creation unit, the simulation of the tire performance by the simulation calculation unit, and the extraction of values of the dimensional parameters by the parameter extraction unit, and machine learning the relationship between the calculation results of the simulation of the tire performance and the extracted dimensional parameters into a prediction model;
a shape determination unit that determines an optimal tire shape for achieving a desired tire performance target value by calculating values of the dimensional parameters using the machine-learned predictive model; and
A tire shape determining device comprising: A tire shape determination device that determines a tire shape using a first tire shape creation system that creates a tire shape using a plurality of dimensional parameters of a tire that can be actually measured with respect to a tire shape indicating a contour line of a tire cross section, and a second tire shape creation system that creates the tire shape using a plurality of non-dimensional parameters related to the tire shape that cannot be actually measured with respect to the tire shape,
a shape creation unit that creates an original tire shape using the non-dimension parameters by the second tire shape creation system;
a parameter extraction unit that extracts values of the dimensional parameters related to a reproduced tire shape obtained by reproducing the original tire shape created by the shape creation unit using the first tire shape creation system;
a model creation unit that creates a tire model having a reproduced tire shape by reproducing the created original tire shape by using the dimensional parameters in the first tire shape creation system;
a simulation calculation unit that simulates tire performance using the tire model;
a prediction model creation unit that prepares a plurality of sets of values of the calculation results of the simulation of the tire performance and values of the dimensional parameters related to the reproduced tire shape by repeatedly performing creation of the tire shape by the shape creation unit, creation of the tire model by the model creation unit, and simulation of the tire performance by the simulation calculation unit, and machine learning the relationship between the calculation results of the simulation of the tire performance and the dimensional parameters into a prediction model;
a shape determination unit that determines an optimal tire shape for achieving a desired tire performance target value by calculating values of the dimensional parameters using the machine-learned predictive model; and
A tire shape determining device comprising:

Images