JP7775745B2 - Method and device for predicting dynamic tire characteristics - Google Patents

Description

This disclosure relates to a method and device for predicting tire dynamic characteristics.

Patent Document 1 below describes a simulation method for evaluating tire traction performance. This method includes a simulation step of rolling a tire model on a road surface model, and a step of acquiring physical quantities related to traction performance from the rolling tire model.

Japanese Patent Application Laid-Open No. 2017-126272

In recent years, technologies have been proposed that create predictive models from existing training data and perform various predictions. Each training data set contains multiple explanatory variables that affect the prediction.

For example, when predicting tire performance, a large number of design factors (explanatory variables) that affect that performance are required. The problem with a large number of such explanatory variables is that a large amount of training data is required to create a predictive model.

This disclosure was devised in light of the above-mentioned circumstances, and its primary objective is to provide a method for predicting tire dynamic characteristics that can reduce the amount of training data required to create a predictive model.

The present disclosure relates to a method for predicting dynamic tire characteristics, including the steps of: inputting at least one design factor of a plurality of tires into a computer; inputting the static characteristics of the plurality of tires into the computer; inputting the dynamic characteristics of the plurality of tires into the computer; using the design factors, static characteristics, and dynamic characteristics of the plurality of tires as training data, the computer creates a prediction model capable of outputting the dynamic characteristics of the tire to be predicted from the design factors and static characteristics of the tire to be predicted; inputting the design factors and static characteristics of the tire to be predicted into the computer; and inputting the design factors and static characteristics of the tire to be predicted into the prediction model and outputting the dynamic characteristics of the tire to be predicted.

By adopting the above steps, the tire dynamic characteristic prediction method disclosed herein makes it possible to reduce the amount of training data required to create a prediction model.

1 is a block diagram showing a computer (tire dynamic characteristic prediction device) on which a tire dynamic characteristic prediction method is executed. FIG. 1 is a cross-sectional view of a tire. FIG. 3 is a partially enlarged view of FIG. 2. 1 is a flowchart showing a processing procedure of a tire simulation method. 1 is a graph conceptually showing a prediction model (approximate response surface) based on a statistical method. 1 is a conceptual diagram of a machine learning-based prediction model. 10 is a flowchart illustrating an example of a processing procedure of a prediction target data input step. 10 is a flowchart showing a processing procedure of a static characteristic input step. FIG. 2 is a perspective view showing a tire model and a road surface model. FIG. 2 is a cross-sectional view showing a tire model. 1 is a flowchart showing a processing procedure of a method for creating a tire dynamic characteristic prediction model. 10 is a graph showing the relationship between predicted values of cornering power and measured values of cornering power in an embodiment. 10 is a graph showing the relationship between predicted values of cornering power and actual measured values of cornering power in a comparative example.

Embodiments of the present disclosure will be described below with reference to the drawings. It should be understood that the drawings may contain exaggerated representations and representations that differ from the dimensional proportions of the actual structure in order to facilitate understanding of the contents of the disclosure. Furthermore, identical or common elements are designated by the same reference numerals throughout each embodiment, and redundant explanations will be omitted. Furthermore, the specific configurations shown in the embodiments and drawings are intended to aid in understanding the contents of the present disclosure, and the present disclosure is not limited to the specific configurations shown.

In the tire dynamic characteristic prediction method of this embodiment (hereinafter sometimes simply referred to as the "prediction method"), tire dynamic characteristics are predicted using a computer. Figure 1 is a block diagram showing a computer 1 (tire dynamic characteristic prediction device 1A) on which the tire dynamic characteristic prediction method of this embodiment is executed.

[Device for predicting tire dynamic characteristics]
The computer 1 of this embodiment has an input unit 2 as an input device, an output unit 3 as an output device, and a processing unit 4 that calculates physical quantities of a tire, etc., and is configured as a tire dynamic characteristic prediction device (hereinafter, sometimes simply referred to as a "prediction device") 1A.

[Input section, output section, and processing unit]
The input unit 2 may be, for example, a keyboard or a mouse. The output unit 3 may be, for example, a display device or a printer. The arithmetic processing unit 4 includes a calculation unit (CPU) 4A that performs various calculations, a storage unit 4B that stores data, programs, etc., and a working memory 4C.

[Storage]
The storage unit 4B is a non-volatile information storage device formed of, for example, a magnetic disk, an optical disk, an SSD, etc. The storage unit 4B includes a data unit 5 and a program unit 6.

[Data section]
The data unit 5 of this embodiment is for storing data and the like necessary for executing the prediction method. The data unit 5 of this embodiment includes a first memory unit 5A, a second memory unit 5B, a third memory unit 5C, and a fourth memory unit 5D. The data unit 5 further includes a fifth memory unit 5E, a sixth memory unit 5F, a seventh memory unit 5G, and an eighth memory unit 5H. Note that the data unit 5 is not limited to this configuration, and may include other memory units, or some of these memory units may be omitted, as necessary.

The first memory unit 5A is for storing the design factors of multiple tires, each of which has at least one different design factor. The second memory unit 5B is for storing the static characteristics of multiple tires. The third memory unit 5C is for storing the dynamic characteristics of multiple tires. The fourth memory unit 5D is for storing the design factors and static characteristics of the tire to be predicted.

The fifth memory unit 5E is for storing data necessary for the simulation described below. The data necessary for the simulation includes, for example, information about the road surface on which the tire will travel, the boundary conditions of the simulation, and the termination conditions of the simulation. The sixth memory unit 5F is for storing the calculation results calculated in the simulation. The seventh memory unit 5G is for storing the created prediction model. The eighth memory unit 5H is for storing the dynamic characteristics of the tire to be predicted.

Program unit 6 is a program (application) required to execute the prediction method. Program unit 6 is executed by calculation unit 4A.

The program unit 6 of this embodiment includes a prediction model creation unit 6A and a dynamic characteristic output unit 6B. Furthermore, the program unit 6 of this embodiment includes a design factor input unit 6C, a static characteristic input unit 6D, a dynamic characteristic input unit 6E, a prediction target data input unit 6F, a simulation calculation unit 6G, and an evaluation unit 6H. Note that the program unit 6 is not limited to this configuration, and other input units may be included, or some of these may be omitted, as necessary.

The prediction model creation unit 6A is used to create a prediction model capable of outputting the dynamic characteristics of the tire to be predicted from the design factors and static characteristics of the tire to be predicted. The dynamic characteristics output unit 6B is used to output the dynamic characteristics of the tire to be predicted.

The design factor input unit 6C is used to input the design factors of multiple tires. The static characteristic input unit 6D is used to input the static characteristics of multiple tires. The dynamic characteristic input unit 6E is used to input the dynamic characteristics of multiple tires. The prediction target data input unit 6F is used to input the design factors and static characteristics of the tire to be predicted. The simulation calculation unit 6G is used to output the static characteristics of the tire to be predicted through simulation. The evaluation unit 6H is used to determine whether the dynamic characteristics of the tire to be predicted are good. The functions of each program unit 6 will be explained in the respective steps of the simulation method described below.

[tire]
2 is a cross-sectional view of the tire 11 of this embodiment. The tire 11 of this embodiment is, for example, a pneumatic tire for a passenger car. However, the tire 11 is not limited to this form and may be, for example, a pneumatic tire for heavy loads.

The tire 11 of this embodiment is provided with a carcass 16 that extends from the tread portion 12 through the sidewall portion 13 to the bead core 15 of the bead portion 14, and a belt layer 17 and a band layer 18 that are disposed radially outward of the carcass 16 and within the tread portion 12.

In this embodiment, the tread portion 12 is provided with tread rubber 19 arranged radially outward of the band layer 18. The tread rubber 19 is composed of a base rubber 19A and a cap rubber 19B arranged radially outward of the base rubber 19A. The sidewall portion 13 is provided with sidewall rubber 20 arranged axially outward of the carcass 16. The bead portion 14 is provided with clinch rubber 21 that contacts the rim 26 when the tire 11 is mounted on the rim 26, and a bead apex rubber 22 that extends radially outward from the bead core 15.

Figure 3 is a partially enlarged view of Figure 2. The tread portion 12 of this embodiment includes a crown region 23A including the tire equator C, shoulder regions 23C including the tread ground-contact edge 12t, and a middle region 23B disposed between the crown region 23A and the shoulder regions 23C. However, the tread portion 12 is not limited to this configuration.

As shown in Figure 2, the tread edge 12t is identified as the axially outermost end of the contact surface 12S of the tread portion 12 when a normal load is applied to a tire 11 in a normal state and the tire is in contact with the ground at a flat surface with a camber angle of 0°. In the normal state, the axial distance between a pair of tread edges 12t, 12t (the other tread edge 12t is omitted in Figure 2) is defined as the tread width TW.

The normal state refers to a state in which the tire 11 is mounted on a normal rim (rim 26), inflated to the normal internal pressure, and no load is applied. Unless otherwise specified, the dimensions of each tire part are shown as values measured in the normal state. Note that the dimensions of each tire part are assumed to allow for normal tolerances inherent in rubber molded products.

A "genuine rim" is a rim specified for each tire by the standard system that includes the standard on which tire 11 is based. Therefore, a genuine rim is, for example, a "standard rim" for JATMA, a "design rim" for TRA, or a "measuring rim" for ETRTO.

"Normal internal pressure" refers to the air pressure specified for each tire by each standard in the standard system, including the standard on which tire 11 is based. Therefore, normal internal pressure is, for example, "maximum air pressure" for JATMA, the maximum value listed in the "TIRE LOAD LIMITS AT VARIOUS COLD INFLATION PRESSURES" table for TRA, and "INFLATION PRESSURE" for ETRTO.

"Normal load" refers to the load specified for each tire in the standard system, including the standard on which tire 11 is based. Therefore, normal load is, for example, "maximum load capacity" for JATMA, the maximum value listed in the "TIRE LOAD LIMITS AT VARIOUS COLD INFLATION PRESSURES" table for TRA, and "LOAD CAPACITY" for ETRTO.

[Tire design factors]
The tire 11 is manufactured based on predetermined design parameters. As shown in Fig. 3, the design parameters in this embodiment include a first radius TR1 of the crown region 23A, a second radius TR2 of the middle region 23B, and a third radius TR3 of the shoulder region 23C. The design parameters also include a fourth radius TR4 of the buttress region 23D between the tread ground-contact edge 12t and the sidewall portion 13.

The design factors in this embodiment include the distance L1 of the first boundary P1, which is the boundary between the crown region 23A and the middle region 23B, from the tire equator C. Furthermore, the design factors include the distance L2 of the second boundary P2, which is the boundary between the middle region 23B and the shoulder region 23C, from the tire equator C. Furthermore, the design factors include the distance L3 of the third boundary P3, which is the boundary between the shoulder region 23C and the buttress region 23D, from the tire equator C. Furthermore, the design factors include the curvature of the outer surface of the sidewall portion 13.

The design factors of this embodiment include the complex modulus E* of the cap rubber 19B, the complex modulus E* of the base rubber 19A, the complex modulus E* of the sidewall rubber 20, and the complex modulus E* of the bead apex rubber 22 (shown in FIG. 2). Each complex modulus E* is measured in accordance with the provisions of JIS K6394 using a dynamic viscoelasticity measuring device (Iplexer series) manufactured by GABO under the following measurement conditions:
Frequency: 10Hz
Initial strain: 5%
Dynamic strain: ±1%
Temperature: 30℃
Deformation mode: tension

The design factors for this embodiment include the angle of the band cords (not shown) arranged in the circumferential direction of the tire that make up the band layer 18 shown in Figure 2 relative to the tire circumferential direction, and the material of the band cords. Furthermore, the design factors include the angle of the belt cords (not shown) that make up the belt layer 17 relative to the tire circumferential direction, the axial width W1 of the belt layer 17, and the radial length L4 of the bead apex rubber 22.

As shown in FIG. 3, the design factors of this embodiment include a first gauge G1, a second gauge G2, a third gauge G3, and a fourth gauge G4. The first gauge G1 is the maximum thickness of the tread rubber 19 at the tire equator C. The second gauge G2 is the maximum thickness of the tread rubber 19 at the first boundary P1. The third gauge G3 is the maximum thickness of the tread rubber 19 at the third boundary P3. The fourth gauge G4 is the maximum rubber thickness of the sidewall rubber 20 between the buttress region 23D and the fourth boundary P4 of the sidewall portion 13.

The design factors of this embodiment include the fifth gauge G5, sixth gauge G6, and seventh gauge G7. As shown in FIG. 2, the fifth gauge G5 is the thickness of the sidewall rubber 20 at the tire's maximum width position P5. The tire's maximum width position P5 is the position where the carcass 16 is at its outermost axial position. The sixth gauge G6 is the maximum width of the clinch rubber 21. As shown in FIG. 3, the seventh gauge G7 is the maximum thickness of the tread rubber 19 at the second boundary P2.

The design factors for this embodiment include the rim width W2 shown in Figure 2, the internal pressure of the tire, and the load applied to the tire. The rim width W2 is the rim width of the standard rim (rim 26) of the tire 11. The internal pressure is the standard internal pressure of the tire 11. The load is the standard load of the tire 11.

The above design factors are examples of design factors for tire 11, and other design factors may also be included.

[Method for predicting tire dynamic characteristics (first embodiment)]
Next, a prediction method of this embodiment will be described. In this embodiment, for a plurality of tires 11 having different design factors, at least one of which is different, a prediction model is created based on training data including the design factors of these tires 11, and the dynamic characteristics of the tire 11A to be predicted are predicted. The dynamic characteristics indicate the characteristics of the tire 11 during running, and examples thereof include cornering force and cornering power.

In general, predicting the dynamic characteristics of tire 11A requires training data that includes numerous design factors (explanatory variables) that affect those dynamic characteristics. Examples of numerous design factors include those listed above. The problem is that the greater the number of such explanatory variables, the greater the amount of training data required to create a prediction model.

Furthermore, as the amount of training data increases, it takes a lot of time to create a prediction model, which means that it takes a lot of time to predict the performance of the tire 11A being predicted. Furthermore, there is the problem of over-learning, which can lead to significant deviations from the trends originally suggested by the training data.

As a result of extensive research, the present inventors have discovered that among the design factors (explanatory variables) necessary for predicting the dynamic characteristics of a tire 11, several design factors can be replaced by a single static characteristic. The static characteristic indicates the characteristics of the tire 11 when stationary (when only a load is applied). The static characteristic includes, for example, parameters related to the contact shape of the tread portion 12 and parameters related to springs.

In the prediction method of this embodiment, the explanatory variables of the prediction model and training data include the static characteristics of the tire 11. This reduces the number of explanatory variables (design factors) and the amount of training data required to create a prediction model with high prediction accuracy. Figure 4 is a flowchart showing the processing steps of the tire simulation method of this embodiment.

[Enter multiple tire design factors]
In the prediction method of this embodiment, first, for a plurality of tires 11 (shown in FIG. 2) each having at least one different tire design factor, the design factors are input into the computer 1 (shown in FIG. 1) (step S1).

In step S1 of this embodiment, first, as shown in FIG. 1, the design factor input unit 6C is loaded into the working memory 4C. The design factor input unit 6C is then executed by the calculation unit 4A, causing the computer 1 to function as a means for inputting the design factors of each of multiple tires 11 (shown in FIGS. 2 and 3).

In step S1 of this embodiment, the design factors are identified for multiple tires 11 that differ in at least one of the above design factors. To identify the design factors, for example, design data (CAD data) for each tire 11 is used, but measurement results of the actual tire 11 may also be used. These design factors are used as explanatory variables for the training data of the prediction model. Furthermore, from the perspective of creating a prediction model 31 that can accurately predict dynamic characteristics, the multiple tires 11 may be in the same category as the tire 11A to be predicted (e.g., pneumatic tires for passenger cars).

As described above, in the prediction method of this embodiment, the explanatory variables of the training data (prediction model) include static characteristics of the tire 11, thereby reducing the number of explanatory variables. When the static characteristics include parameters related to the contact shape of the tread portion 12 and parameters related to springs, each static characteristic is closely related to multiple of the design factors described above and can replace those design factors.

Substitutable design factors include, for example, the first radius TR1, second radius TR2, third radius TR3, fourth radius TR4, distance L1 of the first boundary P1, distance L2 of the second boundary P2, distance L3 of the third boundary P3, and the curvature of the outer surface of the sidewall portion 13, all shown in Figure 3. Other substitutable design factors include the complex modulus E* of the base rubber 19A, the complex modulus E* of the sidewall rubber 20, the complex modulus E* of the bead apex rubber 22 (shown in Figure 2), the angle of the band cord (not shown), and the material of the band cord. Other substitutable design factors include the length L4 of the bead apex rubber 22, the second gauge G2, the fourth gauge G4 to the seventh gauge G7, the rim width W2, the internal pressure, and the load, all of which are shown in Figures 2 and 3. All of these design factors affect the tire's shape and spring characteristics. Therefore, these design factors are closely related (correlated) to static characteristics and can be substituted with static characteristics.

On the other hand, the complex modulus E* of the cap rubber 19B (shown in FIG. 3), the angle of the belt cord (not shown), the width W1 of the belt layer 17 (shown in FIG. 2), and the first and third gauges G1 and G3 shown in FIG. 3 are all factors that significantly affect the rigidity of the tire's contact patch during rolling. Therefore, these design factors cannot be replaced by static characteristics alone. For this reason, in this embodiment, the complex modulus E* of the cap rubber 19B, the angle of the belt cord, the width W1 of the belt layer 17, the first and third gauges G1 and G3 are each stored in the first memory unit 5A (computer 1) shown in FIG. 1 as design factors of multiple tires 11.

[Enter static properties of multiple tires]
Next, in the prediction method of this embodiment, the static characteristics of a plurality of tires 11 (shown in FIG. 2) are input to the computer 1 (shown in FIG. 1) (step S2).

In step S2 of this embodiment, first, as shown in FIG. 1, the static characteristic input unit 6D is loaded into the working memory 4C. Then, the static characteristic input unit 6D is executed by the calculation unit 4A, causing the computer 1 to function as a means for inputting the static characteristics of each of multiple tires 11 (shown in FIG. 2).

As mentioned above, static characteristics include parameters related to the contact shape of the tread portion 12 and parameters related to springs.

Parameters for the contact patch shape of the tread portion 12 (shown in Figure 2) include, for example, the axial width of the contact patch (not shown) (hereinafter sometimes simply referred to as "contact patch width"), the circumferential length of the contact patch (hereinafter sometimes simply referred to as "contact patch length"), and the contact patch area. The contact patch width, contact patch length, and contact patch area may have a zero slip angle or may have a slip angle. Furthermore, the contact patch width, contact patch length, and contact patch area are measured with the tire 11 in a normal state and a normal load applied. Note that the parameters related to the contact patch shape are not limited to the contact patch width, contact patch length, and contact patch area, and can be set as appropriate as long as they relate to the contact patch shape.

Spring-related parameters include, for example, the vertical spring constant of tire 11 (shown in Figure 2) and the lateral spring constant of tire 11. The vertical spring constant can be obtained, for example, using a static testing machine, as the ratio of vertical load to lateral deflection when a vertical load is applied to tire 11 in a normal state. The lateral spring constant can be obtained, for example, using a static testing machine, as the ratio of lateral load to lateral deflection when a lateral load is applied to tire 11 in a normal state. Note that spring-related parameters are not limited to the vertical spring constant and lateral spring constant, and can be set as appropriate as long as they relate to the springs of tire 11.

In step S2 of this embodiment, the lateral spring constant, contact patch width, and contact patch length are acquired as static characteristics of multiple tires 11. These static characteristics are stored in the second memory unit 5B (computer 1) shown in Figure 1.

[Enter static properties of multiple tires]
Next, in the prediction method of this embodiment, the dynamic characteristics of a plurality of tires 11 (shown in FIG. 2) are input to the computer 1 (shown in FIG. 1) (step S3).

In step S3 of this embodiment, first, as shown in FIG. 1, the dynamic characteristics input unit 6E is loaded into the working memory 4C. The dynamic characteristics input unit 6E is then executed by the calculation unit 4A, causing the computer 1 to function as a means for inputting the dynamic characteristics of each of multiple tires 11 (shown in FIG. 2).

As mentioned above, dynamic characteristics indicate the characteristics of the tire 11 while it is running, and include, for example, cornering force and cornering power.

Cornering force refers to the component perpendicular to the direction of travel of the axial lateral force generated by the lateral deformation of the tire 11, which is generated by the tire 11 sliding freely at a slip angle α and causing a small rolling resistance. Cornering power, on the other hand, is the slope of the rise of cornering force at a slip angle of 0° (origin), and is typically determined by measuring cornering force at a slip angle of 1°. Cornering force and cornering power can be measured, for example, using a drum testing machine.

In step S3 of this embodiment, cornering power is acquired as the dynamic characteristic of multiple tires 11. The dynamic characteristic (cornering power) is stored in the third memory unit 5C (computer 1).

[Create a predictive model]
Next, in the prediction method of this embodiment, the computer 1 (shown in FIG. 1) creates a prediction model (step S4). In step S4 of this embodiment, the computer 1 uses the design factors, static characteristics, and dynamic characteristics of a plurality of tires 11 (shown in FIG. 2) as training data to create a prediction model that can output the dynamic characteristics of the tire 11A to be predicted from the design factors and static characteristics of the tire 11A to be predicted.

In step S4 of this embodiment, first, as shown in FIG. 1, the design factors of the multiple tires 11 input into the first memory unit 5A and the static characteristics of the multiple tires 11 input into the second memory unit 5B are loaded into the working memory 4C. Next, the dynamic characteristics of the multiple tires 11 input into the third memory unit 5C are loaded into the working memory 4C. Next, the predictive model creation unit 6A is loaded into the working memory 4C. Then, the predictive model creation unit 6A is executed by the calculation unit 4A, causing the computer 1 to function as a means for creating a predictive model.

The prediction model 31 can be created as appropriate, as long as it is capable of outputting the dynamic characteristics of the tire 11A to be predicted from the design factors and static characteristics of the tire 11A to be predicted. In step S4 of this embodiment, the prediction model is created based on statistical techniques or machine learning.

Statistical methods include, for example, Kriging and Bayesian linear regression. A predictive model based on statistical methods creates an approximate response surface (approximate response function).

Figure 5 is a graph conceptually illustrating a prediction model (approximate response surface) 31 based on statistical techniques. The prediction model (approximate response surface) 31 is capable of accurately representing the relationship between the input (in this example, the tire's design factors and static characteristics) and the output (in this example, the tire's dynamic characteristics) even if there is a highly nonlinear relationship.

In this embodiment, training data (design factors, static characteristics, and dynamic characteristics of multiple tires 11) are substituted into commercially available computer software (e.g., MATLAB ("MATLAB" is a registered trademark) by The MathWorks, Inc., or modeFRONTIER by ESTECO, Inc.). This creates a prediction model 31 in which the design factors and static characteristics of the tires 11 are used as explanatory variables and the dynamic characteristics of the tires 11 are used as response variables. By inputting the design factors and static characteristics of an arbitrary tire 11 as explanatory variables, such a prediction model 31 can obtain the dynamic characteristics of the arbitrary tire 11 by complementing them with the design factors, static characteristics, and dynamic characteristics of multiple known tires 11. The obtained dynamic characteristics of the arbitrary tire 11 can then be predicted (output) as response variables.

The machine learning-based prediction model 31 can be generated, for example, by deep learning using artificial intelligence (AI). Figure 6 is a conceptual diagram of the machine learning-based prediction model 31.

The machine learning-based predictive model 31 is defined by an input layer 32, an output layer 33, and an intermediate layer (hidden layer) 34. The input layer 32 includes design factors and static characteristics 32a, 32b, 32c, etc. of multiple tires 11. The output layer 33 includes dynamic characteristics 33a, 33b, 33c, etc. of multiple tires. The intermediate layer 34 is generated by machine learning.

The design factors and static characteristics 32a, 32b, 32c, etc. of the input layer 32 are linked to, for example, any of the dynamic characteristics 33a, 33b, 33c, etc. of the output layer 33. In other words, the design factors and static characteristics 32a, 32b, 32c, etc. of multiple tires 11 are stored in a linked state with any of the dynamic characteristics 33a, 33b, 33c, etc., and can be used as training data for deep learning.

The intermediate layer 34 includes a combination of multiple neurons (nodes) 35 arranged in multiple layers and optimized weighting coefficients 36 (parameters). Each neuron 35 is connected by a weighting coefficient 36. Such an intermediate layer 34 is called a neural network. In other words, the prediction model 31 of this embodiment includes a neural network.

The weighting coefficients 36 are learned by adjusting the weighting coefficients 36 of each neuron 35, for example, so as to reduce the difference between the output in response to the input and the true output (teaching data). This type of learning method is called backpropagation. Each neuron 35 and each weighting coefficient 36 are determined. This creates a predictive model 31 that uses the design factors and static characteristics of the tire 11 as explanatory variables and the dynamic characteristics of the tire 11 as the objective variable. By inputting the design factors and static characteristics of an arbitrary tire 11 as explanatory variables, this predictive model 31 can predict (output) the dynamic characteristics of the arbitrary tire 11 as the objective variable via the intermediate layer (neural network) 34. Machine learning software (e.g., modeFRONTIER by ESTECO) is used to create the machine learning-based predictive model 31.

In this embodiment, static characteristics are included in the explanatory variables of the prediction model 31, which allows for a reduction in the number of design factors (explanatory variables). As a result, in this embodiment, it is possible to reduce the amount of training data required to create a prediction model 31 with high prediction accuracy. Furthermore, by reducing the amount of training data, it is possible to shorten the time required to create the prediction model 31. Furthermore, by reducing the amount of training data, it is possible to prevent over-learning, which occurs when the training data deviates significantly from the trends originally suggested, and it becomes possible to predict the dynamic characteristics of the tire 11 with high accuracy.

In step S4 of this embodiment, a prediction model 31 (shown in FIG. 5) is created based on a statistical method (Kriging), but a prediction model 31 (shown in FIG. 6) may also be created based on machine learning. The prediction model 31 is stored in the seventh memory unit 5G (computer 1) shown in FIG. 1.

[Enter the design factors and static characteristics of the tire to be predicted]
Next, in the prediction method of this embodiment, design factors and static characteristics of the tire 11A (shown in FIG. 2) to be predicted are input to the computer 1 (shown in FIG. 1) (prediction target data input step S5).

In the prediction target data input step S5 of this embodiment, first, as shown in FIG. 1, the prediction target data input unit 6F is loaded into the working memory 4C. Then, the prediction target data input unit 6F is executed by the calculation unit 4A, causing the computer 1 to function as a means for inputting the design factors and static characteristics of the tire to be predicted. Figure 7 is a flowchart showing an example of the processing procedure for the prediction target data input step S5.

[Enter the design factors of the tire to be predicted]
In the prediction target data input step S5 of this embodiment, first, design factors of the prediction target tire 11A (shown in FIG. 2) are input to the computer 1 (shown in FIG. 1) (step S51). The prediction target tire 11A is not limited to a tire that is actually manufactured, and may be a tire 11 in the design stage (unmanufactured) as long as the dynamic characteristics can be predicted.

In step S51 of this embodiment, the same design factors of the tire 11A to be predicted as the design factors of the multiple tires 11 input in step S1 are input. The design factors of the tire 11A to be predicted in this embodiment include the complex modulus E* of the cap rubber 19B shown in Figures 2 and 3, the angle of the belt cord (not shown), the width W1 of the belt layer 17, the first gauge G1, and the third gauge G3. These design factors are input into the fourth memory unit 5D (computer 1) shown in Figure 1.

[Enter static tire characteristics to be predicted]
Next, in the prediction target data input step S5 of this embodiment, the static characteristics of the tire 11A to be predicted are input to the computer 1 (shown in FIG. 1) (static characteristic input step S52). In the static characteristic input step S52 of this embodiment, of the static characteristics of the tire 11A to be predicted, those that are the same as the static characteristics of the multiple tires 11 input in step S2 are input to the fourth storage unit 5D (computer 1). The static characteristics of the tire 11A to be predicted in this embodiment include the lateral spring constant, contact patch width, and contact patch length.

If the tire 11A to be predicted actually exists (has been manufactured), the static characteristics can be acquired based on the procedure described above in step S2 (shown in FIG. 4) of inputting the static characteristics of multiple tires 11. On the other hand, if the tire 11A to be predicted is still in the design stage (unmanufactured), the static characteristics cannot be acquired directly. For this reason, the static characteristic input step S52 in this embodiment includes a step of acquiring the static characteristics of the tire 11A to be predicted through simulation using a computer 1 (shown in FIG. 1).

In the static characteristic input step S52 of this embodiment, first, as shown in FIG. 1, the simulation calculation unit 6G is loaded into the working memory 4C. Then, the simulation calculation unit 6G is executed by the calculation unit 4A, causing the computer 1 to function as a means for obtaining the static characteristics of the tire 11A to be predicted through simulation. FIG. 8 is a flowchart showing the processing steps of the static characteristic input step S52.

[Enter tire model]
In the static characteristic input step S52 of this embodiment, first, a tire model that models the tire 11A (shown in FIG. 2) to be predicted is input to the computer 1 (shown in FIG. 1) (step S521). In step S521 of this embodiment, the simulation calculation unit 6G (shown in FIG. 1) models the tire model 24 based on the design factors of the tire 11A to be predicted input in step S51. The modeling of the tire model 24 is performed using the same procedure as in Patent Document 1.

Figure 9 is a perspective view showing a tire model 24 and a road surface model 25. Figure 10 is a cross-sectional view showing the tire model 24. Note that the tread pattern and the element F(i) shown in Figure 10 are omitted from Figure 9.

In step S521 of this embodiment, as shown in FIG. 10, the tire 11A (shown in FIG. 2) to be predicted is discretized (modeled) using a finite number of elements F(i) (i = 1, 2, ...). As a result, a tire model 24 is set in step S521.

Element F(i) can be handled using a numerical analysis method. Examples of numerical analysis methods that can be used include the finite element method, finite volume method, finite difference method, or boundary element method (in this embodiment, the finite element method). Element F(i) may be, for example, a three-dimensional tetrahedral solid element, pentahedral solid element, or hexahedral solid element.

Each element F(i) is composed of multiple nodes 37. Numerical data such as the element number, node 37 number, node 37 coordinate values, and material properties (e.g., density, Young's modulus, damping coefficient, thermal conductivity, and heat transfer coefficient) are defined for each element F(i). The tire model 24 is input into the sixth memory unit 5F (computer 1) shown in Figure 1.

[Enter road surface model]
Next, in a static characteristic input step S52 of this embodiment, a road surface model 25 (shown in FIG. 9) that models the road surface is input to the computer 1 (shown in FIG. 1) (step S522). In step S522 of this embodiment, the simulation calculation unit 6G (shown in FIG. 1) models the road surface model 25 based on information about the road surface (e.g., contour data, etc.) that has been input to the fifth storage unit 5E (shown in FIG. 1). The modeling of the road surface model 25 is performed based on the same procedure as in Patent Document 1.

As shown in FIG. 3, in step S522 of this embodiment, based on information about the road surface (not shown), the road surface is discretized using a finite number of elements G(i) (i = 1, 2, ...) that can be handled by a numerical analysis method (in this embodiment, the finite element method). As a result, in step S522, a road surface model 25 that models the road surface is set. The elements G(i) are defined as rigid plane elements that are defined to be non-deformable. The elements G(i) have multiple nodes 38. Furthermore, the elements G(i) are defined with numerical data such as element numbers and coordinate values of the nodes 38. The road surface model 25 is input to the sixth memory unit 5F (computer 1) shown in FIG. 1.

[Define boundary conditions]
Next, in the static characteristic input step S52 of this embodiment, boundary conditions for the simulation are defined in the computer 1 (shown in FIG. 1) (step S523). The boundary conditions include, for example, the internal pressure condition of the tire model 24, the applied load condition L, the camber angle, the slip angle, and the friction coefficient between the tire model 24 and the road surface model 25, which are shown in FIGS. 9 and 10. These boundary conditions are stored in the fifth storage unit 5E (computer 1) shown in FIG. 1 based on the specifications of the tire 11A to be predicted, etc.

[Calculating tire model after internal pressure filling]
Next, in the static characteristic input step S52 of this embodiment, the computer 1 (shown in FIG. 1) calculates the tire model 24 after inflation with internal pressure (step S524). In step S524 of this embodiment, the simulation calculation unit 6G (shown in FIG. 1) calculates the tire model 24 after inflation with internal pressure based on the internal pressure conditions (e.g., normal internal pressure) input to the fifth storage unit 5E (shown in FIG. 1).

In step S524, first, as shown in FIG. 10, the bead portions 24c, 24c of the tire model 24 are constrained by a rim model 28 that models the rim 26 (shown in FIG. 2). Furthermore, in step S524, the deformation of the tire model 24 is calculated based on a uniformly distributed load w that corresponds to the internal pressure condition. This allows the tire model 24 after internal pressure inflation to be calculated.

To calculate the deformation of the tire model 24, a mass matrix, stiffness matrix, and damping matrix are created for each element F(i) based on the shape and material properties of each element F(i). These matrices are then combined to create a matrix for the entire system. Computer 1 (shown in Figure 1) then applies the various conditions described above to create equations of motion, which are used to calculate the deformation of the tire model 24 for each unit time T(x) (x = 0, 1, ...) of the simulation. Such deformation calculations (including rolling calculations, etc., described below) can be performed using commercially available finite element analysis application software, such as LS-DYNA by LSTC. The unit time T(x) can be set appropriately depending on the required simulation accuracy.

[Calculating the tire model after loading]
Next, in the static characteristic input step S52 of this embodiment, the computer 1 (shown in FIG. 1) calculates the tire model 24 after loading (step S525). In this embodiment, the simulation calculation unit 6G (shown in FIG. 1) calculates the tire model 24 after loading (shown in FIG. 9) based on the applied load condition L (normal load), slip angle, camber angle, and friction coefficient input to the fifth storage unit 5E (shown in FIG. 1).

In step S525, as shown in FIG. 9, the contact between the tire model 24 (shown in FIG. 10) after internal pressure inflation and the road surface model 25 is calculated. Next, in step S525, the deformation of the tire model 24 is calculated based on the applied load condition L, slip angle (not shown), camber angle (not shown), and friction coefficient. The applied load condition L is set on the rotation axis 45 of the tire model 24. As a result, in step S525, the tire model 24 after the applied load is calculated, in contact with the road surface model 25.

Calculate static properties
Next, in the static characteristic input step S52 of this embodiment, the computer 1 (shown in FIG. 1) calculates the static characteristics of the tire model 24 (step S526). In this embodiment, the simulation calculation unit 6G (shown in FIG. 1) executed by the calculation unit 4A calculates the static characteristics of the tire model 24 to be predicted based on the tire model 24 after internal pressure filling (shown in FIG. 10) and the tire model 24 after load application (shown in FIG. 9).

In step S526 of this embodiment, the contact patch 24S in contact with the road surface model 25 is identified based on the loaded tire model 24 shown in FIG. 9, and the contact width, contact length, and contact area are calculated. Furthermore, in this embodiment, the vertical spring constant and lateral spring constant can be calculated by applying a vertical load or lateral load to the inflated tire model 24 shown in FIG. 10. The lateral spring constant, contact width, and contact length are obtained as static characteristics of the tire 11A to be predicted. The static characteristics are stored in the fourth memory unit 5D (computer 1) shown in FIG. 1.

[Output the dynamic characteristics of the tire to be predicted]
Next, in the prediction method of this embodiment, the computer 1 (shown in FIG. 1) outputs the dynamic characteristics of the tire 11A (shown in FIG. 2) to be predicted (step S6). In step S6 of this embodiment, the design factors and static characteristics of the tire 11A to be predicted are input to a prediction model 31 (shown in FIG. 5), and the dynamic characteristics of the tire 11A to be predicted (cornering power in this example) are output.

In step S6 of this embodiment, first, as shown in FIG. 1, the prediction model 31 (shown in FIG. 5) input into the seventh memory unit 5G is loaded into the working memory 4C. Next, the design factors and static characteristics of the tire 11A (shown in FIG. 2) to be predicted input into the fourth memory unit 5D are loaded into the working memory 4C. Next, in step S6 of this embodiment, the dynamic characteristic output unit 6B is loaded into the working memory 4C. Then, the dynamic characteristic output unit 6B is executed by the calculation unit 4A, causing the computer 1 to function as a means for outputting the dynamic characteristics of the tire 11A to be predicted.

In step S6 of this embodiment, the design factors and static characteristics of the tire 11A (shown in FIG. 2) to be predicted are input as explanatory variables to the prediction model 31 (shown in FIG. 5). As a result, in step S6, the dynamic characteristics of the tire 11A to be predicted (in this example, cornering power) are output as the objective variable, complemented by the design factors, static characteristics, and dynamic characteristics of multiple known tires 11 (shown in FIG. 2). The dynamic characteristics of the tire 11A to be predicted are output from the output unit 3 (shown in FIG. 1), such as a display device or printer. Furthermore, in this embodiment, the dynamic characteristics of the tire 11A to be predicted are input to the eighth memory unit 5H (shown in FIG. 1).

[Functions of the method and device for predicting tire dynamic characteristics]
In the prediction method (prediction device 1A) of this embodiment, the predictor variables of the prediction model 31 (shown in FIG. 5 ) include static characteristics, thereby reducing the number of explanatory variables (design factors). This improves the prediction accuracy of the dynamic performance of the prediction model 31 (maintaining high robustness against unknown data different from the training data), while reducing the amount of training data required to create the prediction model 31 with high prediction accuracy. Furthermore, reducing the amount of training data shortens the time required to create the prediction model 31, allowing the dynamic characteristics (in this example, cornering power) of the tire 11A (shown in FIG. 2 ) to be predicted to be output in a short time. Furthermore, reducing the amount of training data prevents overlearning, which can result in the results significantly deviating from the trends originally suggested by the training data, thereby enabling the dynamic characteristics of the tire 11A to be predicted to be predicted with high accuracy.

Furthermore, in the prediction method of this embodiment, the static characteristics of the tire 11A (shown in Figure 2) to be predicted are obtained by simulation using a computer 1 (shown in Figure 1), so there is no need to actually manufacture the tire 11A to be predicted. Therefore, the prediction method of this embodiment can output the static and dynamic characteristics of the tire 11A to be predicted in a short time and at low cost.

[Evaluating the dynamic characteristics of the tire to be predicted]
Next, in the prediction method of this embodiment, the computer 1 (shown in FIG. 1) evaluates whether the dynamic characteristics of the tire 11A (shown in FIG. 2) to be predicted are good or not (step S7). As shown in FIG. 1, in step S7 of this embodiment, first, the dynamic characteristics of the tire 11A to be predicted, which are input into the eighth storage unit 5H, and the evaluation unit 6H are read into the working memory 4C. Then, the evaluation unit 6H is executed by the calculation unit 4A, causing the computer 1 to function as a means for evaluating the dynamic characteristics of the tire 11A to be predicted.

Whether the dynamic characteristics are good or not can be evaluated as appropriate. In this embodiment, the dynamic characteristics are determined to be good when the dynamic characteristics (cornering power in this example) are equal to or greater than a predetermined threshold. The threshold can be set as appropriate depending on the cornering performance, etc., required for the tire 11A (shown in Figure 2) being predicted.

If it is determined in step S7 that the dynamic characteristics of the tire 11A (shown in FIG. 2) to be predicted are good ("Yes" in step S7), the tire 11 is manufactured based on the design factors of the tire 11A to be predicted (step S8). On the other hand, if it is determined in step S7 that the dynamic characteristics of the tire 11A to be predicted are not good ("No" in step S7), at least one of the design factors of the tire 11A to be predicted is changed (step S9), and the prediction target data input steps S5 to S7 are performed again. This makes it possible for the prediction method of this embodiment to reliably design and manufacture a tire 11 with good dynamic characteristics.

[Input of static characteristics of multiple tires (second embodiment)]
In step S2 of the above-described embodiment, the static characteristics of the plurality of tires 11 (shown in FIG. 2 ) are acquired based on the measurement results of the actual tires 11, but the present invention is not limited to this. The static characteristics of the plurality of tires 11 may be acquired, for example, by a simulation using a computer 1 (shown in FIG. 1 ).

In step S2 of this embodiment, the static characteristics of multiple tires 11 (shown in FIG. 2) are obtained by simulation based on a procedure similar to the static characteristic input step S52 shown in FIG. 8. As a result, in this embodiment, the static characteristics of multiple tires 11 can be output in a short time and at low cost, without actually manufacturing the multiple tires 11. As a result, in this embodiment, it is possible to shorten the time required to create the prediction model 31 (shown in FIGS. 5 and 6).

[Input of dynamic characteristics of multiple tires (third embodiment)]
In step S3 of the above-described embodiment, the dynamic characteristics of the plurality of tires 11 (shown in FIG. 2 ) are acquired based on the measurement results of the actual tires 11. However, the present invention is not limited to this. The dynamic characteristics of the plurality of tires 11 may be acquired, for example, by a simulation using the computer 1.

In step S3 of this embodiment, tire models 24 (shown in FIG. 9), which are modeled on multiple tires 11 (shown in FIG. 2), are rolled on a road surface model 25 (shown in FIG. 9) to obtain the dynamic characteristics of the multiple tire models 24. The multiple tire models 24 and road surface model 25 are input using the same procedure as in the static characteristic input step S52.

The rolling calculation of the tire model 24 uses the post-load tire model 24 obtained using the same procedure as in the static characteristic input step S52. Then, the rolling calculation of the post-load tire model 24 is performed based on a predetermined traveling speed and turning angle. As a result, in this embodiment, the dynamic characteristics of multiple tires 11 (tire models 24) can be output in a short time and at low cost, without actually manufacturing multiple tires 11. Therefore, in this embodiment, it is possible to shorten the time required to create a predictive model.

[Method for creating a tire dynamic characteristic prediction model]
In the prediction methods of the above-described embodiments, the dynamic characteristics of the tire 11 to be predicted are output using the prediction model 31 (shown in FIGS. 5 and 6 ), but it is also possible to create only the prediction model 31. Fig. 11 is a flowchart showing the processing steps of a method for creating a tire dynamic characteristic prediction model.

In this embodiment of the tire dynamic characteristic prediction model creation method (hereinafter sometimes simply referred to as the "creation method"), a tire 11 dynamic characteristic prediction model 31 (shown in Figures 5 and 6) is created based on steps S1 to S4 of the prediction method shown in Figure 4.

The prediction model 31 can be used for general purposes to predict the dynamic characteristics of the tire 11A (shown in Figure 2) that is the target of prediction. Furthermore, in the prediction method of this embodiment, static characteristics are included in the explanatory variables of the prediction model 31, which reduces the number of design factors (explanatory variables). This makes it possible to reduce the amount of training data required to create a prediction model 31 with high prediction accuracy.

In the creation method of this embodiment, the amount of training data is reduced, which shortens the time required to create the prediction model 31 and prevents over-learning, which can result in a significant deviation from the trends originally suggested by the training data. Therefore, it becomes possible to predict the dynamic characteristics of the tire 11A being predicted with high accuracy.

The above describes in detail a particularly preferred embodiment of the present disclosure, but the present disclosure is not limited to the illustrated embodiment and can be modified and implemented in various ways.

The dynamic characteristics of a tire were predicted based on the processing procedure shown in Figure 4 (Example). In the Example, the design factors, static characteristics, and dynamic characteristics of multiple tires, each of which differed in at least one of the tire design factors, were input into a computer.

Next, in the examples, a prediction model was created that uses the design factors, static characteristics, and dynamic characteristics of multiple tires as training data, and is capable of outputting the dynamic characteristics of the tire to be predicted from the design factors and static characteristics of the tire to be predicted. Then, in the examples, the design factors and static characteristics of multiple tires used as training data were input into the prediction model, and the dynamic performance of those tires was predicted (blind test). Furthermore, in the examples, the design factors and static characteristics of new tires (prediction target data) different from the training data were input into the prediction model, and the dynamic performance of those tires was predicted.

For comparison, a prediction model capable of outputting the dynamic characteristics of a tire to be predicted from the design factors and static characteristics of the tire to be predicted was created using only the design factors and dynamic characteristics of multiple tires as training data (Comparative Example). In the Comparative Example, the design factors of multiple tires used as training data were input into the prediction model, and the dynamic performance of those tires was predicted (blind test). Furthermore, in the Comparative Example, new tire design factors (prediction target data) different from the training data were input into the prediction model, and the dynamic performance of those tires was predicted. Common specifications, etc. are as follows:
Tire sizes: 205/55R16, 235/60R18
Training data: 25 New tires (prediction target data): 100 Dynamic characteristics:
Cornering power Explanatory variables of the example (total: 8):
Design factors:
Complex modulus E* of cap rubber
Belt cord angle
Belt layer width W1
First gauge G1
3rd gauge G3
Static characteristics:
Lateral spring constant
Ground contact width
Contact length Explanatory variables for the comparative example (total: 26):
Design factors:
First radius TR1 to fourth radius TR4
Distance L1 of the first boundary to distance L3 of the third boundary
Outer curvature of the sidewall
Complex modulus of elasticity E* (cap rubber, base rubber, sidewall rubber and
Bead apex rubber)
Angle of band cord relative to the tire circumferential direction
Band cord materials
Angle of belt cord relative to the tire circumferential direction
Belt layer width W1
Length L4 of bead apex rubber in the tire radial direction
1st gauge G1 to 7th gauge G7
Rim width W2
The internal pressure of the tire
Load filled into the tire

Figure 12 is a graph showing the relationship between the predicted cornering power values and the actual measured cornering power values for the example. Figure 13 is a graph showing the relationship between the predicted cornering power values and the actual measured cornering power values for the comparative example. In Figures 12 and 13, the cornering power of multiple tires used as training data is displayed as "□". In addition, the cornering power of new tires (prediction target data) that differ from the training data is displayed as "▲".

The coefficient of determination ^R2 between the predicted value and the measured value of the tire cornering power of the training data was 0.9636 for the Example, while it was 0.9698 for the Comparative Example. As such, since the coefficient of determination ^R2 was approximately the same for the Example and the Comparative Example, the prediction accuracy of the tire cornering power of the training data was similar for the Example and the Comparative Example.

On the other hand, the coefficient of determination ^R2 between the predicted value and the actually measured value of the cornering power of a new tire (prediction target data) different from the training data was 0.9179 for the example, while it was 0.8704 for the comparative example.

Furthermore, the MAE (Mean Absolute Error), which is the average of the absolute values of the difference between the predicted and measured values of the cornering power of the new tire (prediction target data), was 103.28 for the Example, while it was 123.74 for the Comparative Example. Furthermore, the RMSE (Root Mean Squared Error), which is the average of the squared values of the difference between the predicted and measured values of the cornering power of the new tire, was 127.48 for the Example, while it was 160.27 for the Comparative Example.

In the example, the coefficient of determination ^R2 was higher than in the comparative example, and furthermore, the MAE and RMSE were smaller. Therefore, in the example, the prediction accuracy of a new tire (prediction target data) was improved compared to the comparative example. In this way, in the example, by including static characteristics in the explanatory variables of the prediction model, it was possible to reduce the number of explanatory variables and improve the prediction accuracy of the tire's dynamic performance (maintain high robustness against unknown data) with a small amount of training data. Therefore, in the example, it was possible to reduce the amount of training data required to create a prediction model with high prediction accuracy.

[Note]
The present disclosure includes the following aspects.

[Disclosure 1]
1. A method for predicting dynamic properties of a tire, comprising:
a step of inputting at least one design factor of each of a plurality of tires, each of which is different from the others, into a computer;
inputting static characteristics of the plurality of tires into the computer;
inputting dynamic characteristics of the plurality of tires into the computer;
a step in which the computer uses the design factors, the static characteristics, and the dynamic characteristics of the plurality of tires as training data to create a prediction model capable of outputting the dynamic characteristics of the tire to be predicted from the design factors and static characteristics of the tire to be predicted;
A step of inputting design factors and static characteristics of the tire to be predicted into the computer;
and a step in which the computer inputs design factors and static characteristics of the tire to be predicted into the prediction model and outputs dynamic characteristics of the tire to be predicted.
Methods for predicting tire dynamic properties.
[Disclosure 2]
The method for predicting dynamic characteristics of a tire described in the present disclosure 1, further comprising a step of obtaining static characteristics of the tire to be predicted by simulation using the computer.
[Disclosure 3]
The method for predicting dynamic characteristics of a tire according to Disclosure 1 or 2, wherein the static characteristics include parameters related to the contact shape of the tread portion.
[Disclosure 4]
The method for predicting dynamic characteristics of a tire according to any one of Disclosures 1 to 3, wherein the static characteristics include parameters related to springs.
[Disclosure 5]
The method for predicting dynamic characteristics of a tire according to any one of Disclosures 1 to 4, wherein the dynamic characteristics include at least one of cornering force and cornering power.
[Disclosure 6]
The method for predicting dynamic characteristics of a tire according to any one of Disclosures 1 to 5, wherein the step of creating the prediction model creates the prediction model based on a statistical method or machine learning.
[Disclosure 7]
1. An apparatus for predicting dynamic characteristics of a tire, comprising:
a first storage unit for storing at least one design factor of a plurality of tires that are different from each other;
a second storage unit for storing static characteristics of the plurality of tires;
a third storage unit for storing dynamic characteristics of the plurality of tires;
a prediction model creation unit that uses the design factors, the static characteristics, and the dynamic characteristics of the plurality of tires as training data to create a prediction model that can output the dynamic characteristics of the tire to be predicted from the design factors and static characteristics of the tire to be predicted;
a fourth storage unit that stores design factors and static characteristics of the tire to be predicted;
a dynamic characteristic output unit that inputs design factors and static characteristics of the tire to be predicted into the prediction model and outputs dynamic characteristics of the tire to be predicted,
A tire dynamic characteristics prediction device.
[Disclosure 8]
1. A method for creating a predictive model of a tire dynamic characteristic, comprising:
a step of inputting at least one design factor of each of a plurality of tires, each of which is different from the others, into a computer;
inputting static characteristics of the plurality of tires into the computer;
inputting dynamic characteristics of the plurality of tires into the computer;
and a step of generating a prediction model capable of outputting the dynamic characteristics of a tire to be predicted from the design factors and static characteristics of the tire to be predicted, by the computer using the design factors, the static characteristics, and the dynamic characteristics of the plurality of tires as training data.
How to create a tire dynamic property prediction model.

S1 Step of inputting design factors of a plurality of tires S2 Step of inputting static characteristics of a plurality of tires S3 Step of inputting dynamic characteristics of a plurality of tires S4 Step of creating a prediction model S5 Step of inputting design factors and static characteristics of a tire to be predicted S6 Step of outputting dynamic characteristics of a tire to be predicted

Claims (8)

1. A method for predicting dynamic properties of a tire, comprising:
The tire includes a carcass extending from a tread portion through sidewall portions to bead cores of bead portions, and a belt layer disposed radially outward of the carcass and within the tread portion, and also includes a shoulder region including a tread ground-contact edge of the tread portion, and a buttress region between the tread ground-contact edge and the sidewall portion,
the tread portion includes a tread rubber including a base rubber and a cap rubber disposed on the outer side of the base rubber in the tire radial direction,
The method comprises:
a step of inputting at least one design factor of each of a plurality of tires, each of which is different from the others, into a computer;
inputting static characteristics of the plurality of tires into the computer;
inputting dynamic characteristics of the plurality of tires into the computer;
a step in which the computer uses the design factors, the static characteristics, and the dynamic characteristics of the plurality of tires as training data to create a prediction model capable of outputting the dynamic characteristics of the tire to be predicted from the design factors and static characteristics of the tire to be predicted;
A step of inputting design factors and static characteristics of the tire to be predicted into the computer;
and a step in which the computer inputs design factors and static characteristics of the tire to be predicted into the prediction model and outputs dynamic characteristics of the tire to be predicted ,
the design factors include a complex elastic modulus of the cap rubber, an angle of a belt cord constituting the belt layer, a width of the belt layer, a first gauge which is a maximum thickness of the tread rubber at the tire equator, and a third gauge which is a maximum thickness of the tread rubber at a third boundary which is a boundary between the shoulder region and the buttress region,
The static characteristics include a parameter related to the ground contact shape of the tread portion and a parameter related to spring.
Methods for predicting tire dynamic properties.
The method for predicting dynamic characteristics of a tire described in claim 1 further includes a step of obtaining the static characteristics of the tire to be predicted by simulation using the computer. 3. The method for predicting dynamic characteristics of a tire according to claim 1, wherein the static characteristics of the plurality of tires are obtained by a simulation using the computer .
4. The method for predicting dynamic characteristics of a tire according to claim 1, wherein the dynamic characteristics of the plurality of tires are obtained by a simulation using the computer .
A method for predicting dynamic characteristics of a tire according to any one of claims 1 to 4, wherein the dynamic characteristics include at least one of cornering force and cornering power. The method for predicting dynamic characteristics of a tire according to any one of claims 1 to 5, wherein the step of creating the predictive model creates the predictive model based on statistical techniques or machine learning. 1. An apparatus for predicting dynamic characteristics of a tire, comprising:
The tire includes a carcass extending from a tread portion through sidewall portions to bead cores of bead portions, and a belt layer disposed radially outward of the carcass and within the tread portion, and also includes a shoulder region including a tread ground-contact edge of the tread portion, and a buttress region between the tread ground-contact edge and the sidewall portion,
the tread portion includes a tread rubber including a base rubber and a cap rubber disposed on the outer side of the base rubber in the tire radial direction,
The device comprises:
a first storage unit for storing at least one design factor of a plurality of tires that are different from each other;
a second storage unit for storing static characteristics of the plurality of tires;
a third storage unit for storing dynamic characteristics of the plurality of tires;
a prediction model creation unit that uses the design factors, the static characteristics, and the dynamic characteristics of the plurality of tires as training data to create a prediction model that can output the dynamic characteristics of the tire to be predicted from the design factors and static characteristics of the tire to be predicted;
a fourth storage unit that stores design factors and static characteristics of the tire to be predicted;
a dynamic characteristic output unit that inputs design factors and static characteristics of the tire to be predicted into the prediction model and outputs dynamic characteristics of the tire to be predicted ,
the design factors include a complex elastic modulus of the cap rubber, an angle of a belt cord constituting the belt layer, a width of the belt layer, a first gauge which is a maximum thickness of the tread rubber at the tire equator, and a third gauge which is a maximum thickness of the tread rubber at a third boundary which is a boundary between the shoulder region and the buttress region,
The static characteristics include a parameter related to the ground contact shape of the tread portion and a parameter related to spring.
A tire dynamic characteristics prediction device.
1. A method for creating a predictive model of a tire dynamic characteristic, comprising:
The tire includes a carcass extending from a tread portion through sidewall portions to bead cores of bead portions, and a belt layer disposed radially outward of the carcass and within the tread portion, and also includes a shoulder region including a tread ground-contact edge of the tread portion, and a buttress region between the tread ground-contact edge and the sidewall portion,
the tread portion includes a tread rubber including a base rubber and a cap rubber disposed on the outer side of the base rubber in the tire radial direction,
The method comprises:
a step of inputting at least one design factor of each of a plurality of tires, each of which is different from the others, into a computer;
inputting static characteristics of the plurality of tires into the computer;
inputting dynamic characteristics of the plurality of tires into the computer;
and a step of generating a prediction model capable of outputting the dynamic characteristics of a tire to be predicted from the design factors and static characteristics of the tire to be predicted, by the computer using the design factors, the static characteristics, and the dynamic characteristics of the plurality of tires as training data ,
the design factors include a complex elastic modulus of the cap rubber, an angle of a belt cord constituting the belt layer, a width of the belt layer, a first gauge which is a maximum thickness of the tread rubber at the tire equator, and a third gauge which is a maximum thickness of the tread rubber at a third boundary which is a boundary between the shoulder region and the buttress region,
The static characteristics include a parameter related to the ground contact shape of the tread portion and a parameter related to spring.
How to create a tire dynamic property prediction model.