JP2021195038A - Method for learning tire performance prediction model, method for predicting tire performance, system and program - Google Patents

Abstract

To provide a method for learning a tire performance prediction model that predicts a tire performance value on the basis of an image of a tire ground contact area, and to provide a method for predicting tire performance, a system and a program.SOLUTION: The method for learning a tire performance prediction model that is executed by one or more processors includes: a feature amount extraction step of inputting an input image on the basis of a ground contact area image indicating a tire ground contact surface shape into a feature extracting unit to extract a feature amount of the image; and a learning step of making a machine to learn a prediction model so as to output a tire performance value using the extracted feature amount of the extracted image as the input therefor.SELECTED DRAWING: Figure 3

Description

The present disclosure relates to a learning method of a tire performance prediction model, a tire performance prediction method, a system and a program.

The use of machine learning is spreading to various industrial fields, and the number of cases of application to tires of other companies in the same industry is increasing. For example, Patent Document 1 describes that an image showing a tread of a tire is input to a neural network to estimate a wear state or a wear amount.

The inventors of the present disclosure believe that tire performance is manifested in the tread, as tire designers often discuss based on the tread. Therefore, if the tire performance value can be predicted from the contact patch image showing the contact patch shape of the tire that touches the road surface under a predetermined load, it will lead to a reduction in the number of trial trials and tests, which is considered useful. It also provides a clue as to which element of the tread is related to each performance of the tire. However, no specific method for predicting tire performance from the contact patch image of the tire has been proposed.

Japanese Unexamined Patent Publication No. 2019-35626

The present disclosure provides a learning method, a tire performance prediction method, a system and a program of a tire performance prediction model for predicting a tire performance value based on a contact patch image of a tire.

The learning method of the tire performance prediction model of the present disclosure is a method executed by one or a plurality of processors, in which an input image based on a contact patch image representing the tire contact patch shape is input to the feature extraction unit and the feature amount of the image is obtained. It includes a feature amount extraction step for extracting a tire, and a learning step for machine learning a prediction model so as to output a tire performance value by using the feature amount of the extracted image as an input.

The figure which shows the usage mode of the tire performance prediction model learning system and the tire performance prediction system of 1st Embodiment. The block diagram which shows the tire performance prediction model learning system and the tire performance prediction system of 1st Embodiment. The flowchart executed by the tire performance prediction model learning system of 1st Embodiment. The flowchart which executes the tire performance prediction system of 1st Embodiment. Explanatory drawing about the process of trimming a ground plane image and generating an input image. The block diagram which shows the tire performance prediction model learning system and the tire performance prediction system of 2nd Embodiment. A flowchart executed by the tire performance prediction model learning system of the second embodiment. The flowchart which executes the tire performance prediction system of 2nd Embodiment.

<First Embodiment>
Hereinafter, the first embodiment of the present disclosure will be described with reference to the drawings.

FIG. 1 is a diagram showing a usage mode of the tire performance prediction model learning system 1 and the tire performance prediction system 2. As shown in FIG. 1, the tread image 5 is obtained from the test device 3 or the tire tread simulation system 4. The tire performance prediction model learning system 1 machine-learns the prediction model using the teacher data associated with the contact patch image 5 as an input to the prediction model and the correct tire performance value as an output from the prediction model. The input image used for the teacher data is based on the ground plane image. The tire performance prediction system 2 calculates (predicts) and outputs a tire performance value based on the contact patch image 5 to be predicted by using the prediction model constructed by the tire performance prediction model learning system 1.

The tire performance value to be predicted is an actually measured value. The tire performance values used in this embodiment are CP (cornering power), SAP (self-aligning power), CFmax (maximum cornering force), and SAT (self-aligning torque). Of course, the tire performance value can be applied to any other tire performance.

The contact patch image 5 shows the shape of the tire contact patch. The contact patch image 5 of the present embodiment represents the shape of the tire contact patch, and the contact patch Pz in the direction perpendicular to the road surface is represented by color in the case of a color image and by luminance in the case of gray scale. The ground plane image 5 of the present embodiment is an image of a stationary state in which the tire is not rolling, but the image is not limited to this, and may be an image of rolling. The camber angle of the tire can be set to any angle regardless of whether it is in a rolling state or a stationary state. The ground plane image 5 of the present embodiment is an image of a stationary state of the tire, but when the tire is in a rolling state, the slip angle with respect to the traveling direction may be 0 degrees or may be an angle other than 0 degrees. .. The ground plane image 5 of the present embodiment represents only the ground plane pressure Pz in the direction perpendicular to the road surface in a stationary state, but is not limited thereto. For example, when the tire is in a rolling state, the pressure Px along the traveling direction of the tire or the pressure Py along the direction orthogonal to the traveling direction of the tire may be expressed. Note that these coordinate systems are examples and can be changed as appropriate. If the accuracy can be ensured, the contact patch image 5 may represent only the shape of the tire contact patch. The test device 3 touches the tire 30 to be tested on the test road surface 31 under a predetermined load, and photographs the ground contact surface shape with the camera 32 through the transparent road surface on the test road surface 31. The ground contact pressure is measured by an optical method using a transparent plate or the like, or by using a pressure sensor. The test device 3 generates a ground plane image 5 by the above test. Further, the contact patch image 5 may be acquired from the simulation result obtained by the tire contact patch simulation system 4 shown in FIG.

Depending on the mounting method, the tire performance prediction model learning system 1 and the tire performance prediction system 2 are not constructed on the same computer system, and can be operated independently. That is, only the tire performance prediction model learning system 1 may be implemented, or only the tire performance prediction system 2 may be implemented.

[Tire performance prediction model learning system 1]
FIG. 2 is a block diagram showing a tire performance prediction model learning system 1 and a tire performance prediction system 2. As shown in FIG. 2, the tire performance prediction model learning system 1 has a conversion unit 10, a feature extraction unit 11, and a learning unit 12. The conversion unit 10 can be omitted.

Each of these parts 10 to 12 is realized in cooperation with software and hardware by the processor 1a executing the processing routine shown in FIG. 3 which is stored in advance in a computer equipped with a processor 1a, a memory 1b, various interfaces, and the like. Will be done. In the present embodiment, the processor 1a in one device realizes each part, but the present invention is not limited to this. For example, it may be distributed using a network, and a plurality of processors may be configured to execute the processing of each part. That is, one or more processors execute the process.

When the input image 6 based on the ground plane image 5 is input, the feature extraction unit 11 extracts the feature amount of the image. In the configuration in which the conversion unit 10 is not provided, the ground plane image 5 is input to the feature extraction unit 11 as the input image 6. In the configuration provided with the conversion unit 10, the input image 6 output by the conversion unit 10 is input. The feature extraction unit 11 may use any algorithm as long as the feature amount can be extracted from the image. For example, a neural network (for example, Alexnet, GoogleNet, ResNet101), a discrete cosine transform process, an AutoEncoder, a configuration in which a discrete cosine transform process and an AutoEncoder are used in combination can be mentioned.

The learning unit 12 uses the teacher data set D1 (or D2) to machine-learn and construct the prediction model 13 so as to output the tire performance value by inputting the feature amount of the image extracted by the feature extraction unit 11. .. In the teacher data set D1, the ground plane image 5 (ground plane images 1 to N) or the input image 6 (input images 1 to _N ) was associated with the _{tire performance (X 1} , X ₂ , ..., X N) values. It is data. The teacher data set D2 is data in which the input image 6 (input images 1 to N) and the tire performance (X ₁ , X ₂ , ..., X _N ) values are associated with each other. N indicates the number of teacher data. As the prediction model 13, various models such as Gaussian process regression, linear regression, classification tree, random forest, support vector machine, and ensemble tree can be used as long as they are supervised machine learning models.

The conversion unit 10 converts the ground plane image 5 into an input image 6 having a size that can be input to the feature extraction unit 11. In this embodiment, a color or grayscale (8-bit) ground plane image 5 of 875 × 656 pixels is converted into a grayscale (8-bit) input image 6 of 227 × 227 pixels, which is an example. , Not limited to this.

In converting to a size that can be input to the feature extraction unit 11, it is necessary not to destroy the features of the image appearing in the original ground plane image 5. The size of the ground plane affects performance. For example, the contact patch of a tire with a large tire size is large in the image, and conversely, the contact patch of a tire with a small tire size is small in the image. If this is trimmed so that the size of the contact patch in the image after trimming is the same, the performance value should be predicted to be small for tires with small tire sizes, but it is the same as tires with large tire sizes. There is a risk of overestimating the performance. Therefore, since the size of the contact patch shape affects the tire performance value, it is necessary to maintain the scale among a plurality of images.

Therefore, the conversion unit 10 includes a selection unit 10a, a trimming position determining unit 10b, a trimming unit 10c, and a size changing unit 10d. As a premise, all the ground plane images 5 are taken under the same shooting conditions, and the scales are the same. The same shooting condition means that, as shown in FIG. 1, the distance from the camera 32 to the test road surface 31 is the same, and the zoom value of the camera 32 is the same.

The selection unit 10a selects the contact patch image having the largest contact patch shape from the teacher data set D1 in which the contact patch image 5 and the tire performance (X _{1 to N) are associated with each other, which is used for machine learning.} In the case of the ground plane images 50, 51, 52 shown in FIG. 5, the ground plane image 50 has the largest ground plane shape.

As shown in FIG. 5, the trimming position determination unit 10b determines the trimming position P1 in the image based on the ground plane image 50 selected by the selection unit 10a. The larger the contact patch shape in the input image 6, the better the prediction accuracy. Therefore, in the present embodiment, the trimming position P1 is the range in which the minimum rectangle including the ground plane shape or the minimum rectangle is expanded by a predetermined pixel. The trimming position P1 determined by the trimming position determining unit 10b is stored in the memory 1b for use by the tire performance prediction system 2.

The trimming unit 10c trims all the ground plane images 5 (images 1 to N) of the teacher data set D1 using the trimming position P1 determined by the trimming position determining unit 10b to generate trimmed images, respectively.

The size changing unit 10d changes each trimmed image generated by the trimming unit 10c to a size that can be input to the feature extraction unit 11 to generate an input image 6. The resizing unit 10d resizes the trimmed image without changing the aspect ratio. In the example of FIG. 5, the input image 60 is generated from the ground plane image 50 via a trimmed image (not shown). An input image 61 is generated from the ground plane image 51 via a trimmed image (not shown). An input image 62 is generated from the ground plane image 52 via a trimmed image (not shown). By executing such a process, it is possible to prevent the scale of the ground plane shape reflected in each of the input images 60 to 62 from becoming disjointed.

The tire performance that improves the prediction performance by trimming while maintaining the aspect ratio in one image and the relative size (scale) between a plurality of images by the conversion unit 10 is as follows.
Cornering system: CP, SAP, CFmax (maximum cornering force), SAT
Traction system: Braking performance on dry road surface, Braking performance on wet road surface, Braking performance on ice road surface, Braking performance on snowy road surface, Friction performance on ice, Friction performance on snow, Hydroplaning resistance Noise system: Noise performance inside and outside the vehicle, tire unit test Radiation sound performance in rolling resistance Abrasion resistance Heel & toe wear performance and uneven wear performance are affected by the ratio and distribution of areas with high contact patch in the contact patch, the shape of the contact patch, etc., as described above. I don't think there is a need for trimming or resizing with tire size information.

[Tire performance prediction system 2]
As shown in FIG. 2, the tire performance prediction system 2 includes a conversion unit 20, a feature extraction unit 21, and a prediction unit 22. The conversion unit 20 can be omitted as in the tire performance prediction model learning system 1.

When the input image 6 based on the ground plane image 5 is input, the feature extraction unit 21 extracts the feature amount of the image. In the configuration in which the conversion unit 20 is not provided, the ground plane image 5 is input to the feature extraction unit 21 as the input image 6. In the configuration provided with the conversion unit 20, the input image 6 output by the conversion unit 20 is input. The feature extraction unit 21 in the tire performance prediction system 2 has the same configuration as the feature extraction unit 11 in the tire performance prediction model learning system 1.

The prediction unit 22 inputs the feature amount of the image output by the feature extraction unit 21 and outputs the tire performance value by using the prediction model 13 constructed by the tire performance prediction model learning system 1.

As shown in FIG. 2, the conversion unit 20 converts the ground plane image 5 to be predicted into an input image 6 suitable for input to the feature extraction unit 21. The conversion unit 20 includes a trimming unit 20c and a size changing unit 20d. The trimming unit 20c trims the ground plane image 5 using a predetermined trimming position P1 to generate a trimmed image. The trimming position P1 is determined by the trimming position determining unit 10b and stored in the memory 1b. The size changing unit 20d changes the size of the trimmed image generated by the trimming unit 20c to a size that can be input to the feature extraction unit 21, and generates the input image 6. The resizing unit 20d resizes the trimmed image without changing the aspect ratio. The trimming unit 20c has the same configuration as the trimming unit 10c in the tire performance prediction model learning system 1. The size changing unit 20d has the same configuration as the size changing unit 10d in the tire performance prediction model learning system 1.

[Learning method of tire performance prediction model]
A method of learning a tire performance prediction model executed by one or more processors in the tire performance prediction model learning system 1 shown in FIG. 2 will be described with reference to FIG.

First, by executing steps ST1 to ST4, the conversion unit 10 generates an input image 6 based on the ground plane image 5. Specifically, in step ST1, the selection unit 10a is one of the teacher data sets D2 in which the contact patch image 5 (images 1 to N) and the tire performance (X _{1 to N) are associated with each other, which is used in machine learning.} The ground plane image 50 having the largest ground plane shape is selected.
In the next step ST2, the trimming position determination unit 10b determines the trimming position P1 in the image based on the selected ground plane image 50.
In the next step ST3, the trimming unit 10c trims all the ground plane images (1 to N) of the teacher data set D1 using the trimming position P1 determined by the trimming position determining unit 10b, and trims the trimmed images. Generate.
In the next step ST4, the size changing unit 10d changes the size of each trimmed image to a size that can be input to the feature extraction unit 11 without changing the aspect ratio of each trimmed image, and inputs the input image 6 ( 60-62) is generated.
In the next step ST5, the feature extraction unit 11 inputs the input image 6 based on the contact patch image 5 representing the tire contact patch shape to the feature extraction unit 11 and extracts the feature amount of the image.
In the next step ST6, the learning unit 12 machine-learns the prediction model 13 so as to output the tire performance value by inputting the feature amount of the extracted image.

[Tire performance prediction method]
A tire performance prediction method executed by one or more processors in the tire performance prediction system 2 shown in FIG. 2 will be described with reference to FIG.

First, by executing steps ST101 to 102, the conversion unit 20 generates an input image 6 based on the ground plane image 5. Specifically, in step ST101, the trimming unit 20c trims the ground plane image 5 to be predicted using a predetermined trimming position P1 to generate a trimmed image.
In the next step ST102, the size changing unit 20d changes the size of the trimmed image to a size that can be input to the feature extraction unit 21 without changing the aspect ratio of the trimmed image, and generates the input image 6.
In the next step ST103, the feature extraction unit 21 inputs the input image 6 based on the contact patch image 5 representing the tire contact patch shape to the feature extraction unit 21 and extracts the feature amount of the image.
In the next step ST104, the prediction unit 22 corresponds to the feature amount of the extracted image by using the prediction model 13 machine-learned to output the tire performance value by inputting the feature amount of the extracted image. Output the tire performance value.

<Second Embodiment>
Hereinafter, the second embodiment of the present disclosure will be described with reference to the drawings.

As shown in FIG. 6, the tire performance prediction model learning system 1 and the tire performance prediction system 2 of the second embodiment execute the trimming process using the data D3 in which the tire size and the trimming position are previously associated with each other. ,It is configured.

Specifically, as shown in FIGS. 6 and 7, the conversion unit 10 of the tire performance prediction model learning system 1 of the second embodiment has a trimming position specifying unit 10e. The memory 1b stores data D3 in which the tire size and the trimming position are associated with each other.
The operation of the tire performance prediction model learning system 1 is as follows.
First, by executing steps ST2011 to 203, the conversion unit 10 generates an input image 6 based on the ground plane image 5. Specifically, in step ST201, the trimming position specifying unit 10e specifies the trimming position corresponding to the designated tire size based on the data D3 in which the tire size and the trimming position are previously associated with each other.
In the next step ST202, all the ground plane images (1 to N) of the teacher data set D1 are trimmed to generate trimmed images by using the trimming position P1 specified by the trimming position specifying unit 10e.
In the next step ST203, the size changing unit 10d changes the size of each trimmed image to a size that can be input to the feature extraction unit 11 without changing the aspect ratio of each trimmed image, and inputs the input image 6 ( 60-62) is generated.
In the next step ST204, the feature extraction unit 11 inputs the input image 6 based on the contact patch image 5 representing the tire contact patch shape to the feature extraction unit 11 and extracts the feature amount of the image.
In the next step ST205, the learning unit 12 machine-learns the prediction model 13 so as to output the tire performance value by inputting the feature amount of the extracted image.

Specifically, as shown in FIGS. 6 and 8, the conversion unit 20 of the tire performance prediction system 2 of the second embodiment has a trimming position specifying unit 20e. The memory 1b stores data D3 in which the tire size and the trimming position are associated with each other.
The operation of the tire performance prediction system 2 is as follows.
First, by executing steps ST301 to 303, the conversion unit 20 generates an input image 6 based on the ground plane image 5. Specifically, in step ST301, the trimming position specifying unit 20e specifies the trimming position corresponding to the designated tire size based on the data D3 in which the tire size and the trimming position are previously associated with each other.
In the next step ST302, the trimming unit 20c trims the ground plane image 5 to be predicted using the trimming position P1 specified by the trimming position specifying unit 20e to generate a trimmed image.
In the next step ST303, the size changing unit 20d changes the size of the trimmed image to a size that can be input to the feature extraction unit 21 without changing the aspect ratio of the trimmed image, and generates the input image 6.
In the next step ST304, the feature extraction unit 21 inputs the input image 6 based on the contact patch image 5 representing the tire contact patch shape to the feature extraction unit 21 and extracts the feature amount of the image.
In the next step ST305, the prediction unit 22 corresponds to the feature amount of the extracted image by using the prediction model 13 machine-learned to output the tire performance value by inputting the feature amount of the extracted image. Output the tire performance value.

As described above, the learning method of the tire performance prediction model of the first embodiment or the second embodiment is a method executed by one or a plurality of processors, and is an input based on the contact patch image 5 representing the tire contact patch shape. A feature amount extraction step in which the image 6 is input to the feature extraction unit 11 to extract the feature amount of the image, and learning to machine-learn the prediction model 13 so as to output the tire performance value by inputting the feature amount of the extracted image. Including steps.

The tire performance prediction model learning system 1 of the first embodiment or the second embodiment has a feature extraction unit 11 that inputs an input image 6 based on a contact patch image 5 representing a tire contact patch shape and extracts a feature amount of the image. , A learning unit 12 that machine-learns the prediction model 13 so as to output a tire performance value by inputting a feature amount of the extracted image.

As a result, it is possible to provide a prediction model 13 that predicts the tire performance value based on the feature amount extracted from the input image 6 based on the contact patch image 5, and it is possible to know the tire performance value based on the contact patch image 5.

The tire performance prediction method of the first embodiment or the second embodiment is a method executed by one or a plurality of processors, and the input image 6 based on the tire contact surface image 5 representing the tire contact surface shape is input to the feature extraction unit 21. Using the feature amount extraction step of inputting and extracting the feature amount of the image, and the prediction model 13 machine-learned to output the tire performance value by inputting the feature amount of the extracted image, the extracted image Includes a prediction step that outputs the tire performance value corresponding to the feature amount.

The tire performance prediction system 2 of the first embodiment or the second embodiment has a feature extraction unit 21 for inputting an input image 6 based on a contact patch image 5 representing a tire contact patch shape and extracting feature amounts of the image, and an extraction unit. Using the prediction model 13 machine-learned to output the tire performance value by inputting the feature amount of the extracted image, the prediction unit 22 that outputs the tire performance value corresponding to the feature amount of the extracted image. Be prepared.

As a result, the prediction model 13 predicts the tire performance value based on the feature amount extracted from the input image 6 based on the contact patch image 5, so that the tire performance value can be known based on the contact patch image 5.

Although not particularly limited, it is preferable that the contact patch image 5 represents the tire contact patch shape and the contact pressure as in the first embodiment or the second embodiment.

Although not particularly limited, the image generation step includes an image generation step of generating an input image 6 based on the ground plane image 5 as in the learning method of the tire performance prediction model of the first embodiment, and the image generation step is used in the learning step. Of the teacher data set D1 that associates the ground plane image 5 with the tire performance value, the step of selecting the ground plane image 50 having the largest ground plane shape and the trimming position P1 in the image based on the selected ground plane image 50. The step to determine, the step to trim all the ground plane images 5 of the teacher data set D1 using the determined trimming position P1 to generate trimmed images, and the aspect ratio of each trimmed image is unchanged. May include a step of changing the size of each trimmed image to a size that can be input to the feature extraction unit 11 to generate an input image 6.

As a result, since the ground plane image 5 having the largest ground plane shape is selected and the trimming position P1 is determined, the ground plane occupied in the trimmed image can be made as large as possible, and the prediction accuracy by the prediction model 13 can be improved. Become. Nevertheless, since the determined trimming position is used in common for all the contact patch images 5 for trimming, the tire performance value can be correctly predicted in consideration of the difference in the size of the contact patch shape depending on the tire size.

Although not particularly limited, the image generation step includes an image generation step of generating an input image 6 based on the ground plane image 5 as in the tire performance prediction method of the first embodiment, and the image generation step uses a predetermined trimming position. The step of trimming the ground plane image 5 to generate a trimmed image and the input image 6 by changing the size of the trimmed image to a size that can be input to the feature extraction unit 21 without changing the aspect ratio of the trimmed image. And may include.

As a result, since trimming is performed by using a predetermined trimming position in common, it is possible to correctly predict the tire performance value in consideration of the difference in the size of the contact patch shape depending on the tire size.

Although not particularly limited, the image generation step includes an image generation step of generating an input image 6 based on the ground plane image 5 as in the learning method of the tire performance prediction model or the tire performance prediction method of the second embodiment, and the image generation step is a tire. Based on the data D3 in which the size and the trimming position are associated in advance, the step of specifying the trimming position corresponding to the specified tire size and the ground plane image 5 are trimmed using the specified trimming position to generate a trimmed image. And the step of changing the size of the trimmed image to a size that can be input to the feature extraction units 11 and 21 without changing the aspect ratio of the trimmed image to generate the input image 6. good.

As a result, trimming is performed using a common or separate appropriate trimming position according to the tire size, so that the tire performance value can be correctly predicted in consideration of the difference in the size of the contact patch shape depending on the tire size.

The program according to this embodiment is a program that causes a computer to execute the above method.
By executing these programs, it is possible to obtain the effects of the above method.

Although the embodiments of the present disclosure have been described above with reference to the drawings, it should be considered that the specific configuration is not limited to these embodiments. The scope of the present disclosure is set forth not only by the description of the embodiment described above but also by the scope of claims, and further includes all modifications within the meaning and scope equivalent to the scope of claims.

It is possible to adopt the structure adopted in each of the above embodiments in any other embodiment. The specific configuration of each part is not limited to the above-described embodiment, and various modifications can be made without departing from the gist of the present disclosure.

For example, the execution order of each process such as operation, procedure, step, and step in the device, system, program, and method shown in the claims, specification, and drawings may be the output of the previous process. It can be realized in any order unless it is used in processing. Even if the scope of claims, the specification, and the flow in the drawings are explained using "first", "next", etc. for convenience, it does not mean that it is essential to execute in this order. ..

Each part shown in FIGS. 2 and 7 is realized by executing a predetermined program by 1 or a processor, but each part may be configured by a dedicated memory or a dedicated circuit. In the system 1 of the above embodiment, each part is mounted on the processor 1a of one computer, but each part may be distributed and mounted on a plurality of computers or the cloud. That is, the above method may be executed by one or more processors.

The system 1 includes a processor 1a. For example, the processor 1a can be a central processing unit (CPU), a microprocessor, or any other processing unit capable of executing computer-executable instructions. Further, the system 1 includes a memory 1b for storing the data of the system 1. In one example, memory 1b includes a computer storage medium such as RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, DVD or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other. Includes a magnetic storage device, or any other medium that can be used to store desired data and that System 1 can access.

1 ... Tire performance prediction model learning system, 10 ... Conversion unit, 11 ... Feature extraction unit, 12 ... Learning unit, 2 ... Tire performance prediction system, 20 ... Conversion unit, 21 ... Feature extraction unit, 22 ... Prediction unit

Claims (8)

A method executed by one or more processors.
A feature amount extraction step in which an input image based on a contact patch image representing the shape of the tire contact patch is input to the feature extraction unit to extract the feature amount of the image, and
A learning step in which the prediction model is machine-learned so as to output the tire performance value by inputting the feature amount of the extracted image.
How to learn a tire performance prediction model, including. A method executed by one or more processors.
A feature amount extraction step in which an input image based on a contact patch image representing the shape of the tire contact patch is input to the feature extraction unit to extract the feature amount of the image, and
Using a prediction model machine-learned to output the tire performance value by inputting the feature amount of the extracted image, the prediction step of outputting the tire performance value corresponding to the feature amount of the extracted image. ,
Tire performance prediction methods, including. The method according to claim 1 or 2, wherein the contact patch image represents the tire contact patch shape and contact pressure. Including an image generation step of generating the input image based on the ground plane image.
The image generation step is
A step of selecting the contact patch image having the largest contact patch shape from the teacher data set associated with the contact patch image and the tire performance value used in the learning step.
The step of determining the trimming position in the image based on the selected ground plane image, and
A step of trimming the ground plane image of the teacher data set using the determined trimming position to generate a trimmed image, respectively.
A step of changing the size of each trimmed image to a size that can be input to the feature extraction unit without changing the aspect ratio of each trimmed image to generate the input image.
The method according to claim 1. Including an image generation step of generating the input image based on the ground plane image.
The image generation step is
A step of trimming the ground plane image using a predetermined trimming position to generate a trimmed image, and
A step of changing the size of the trimmed image to a size that can be input to the feature extraction unit without changing the aspect ratio of the trimmed image to generate the input image.
2. The method according to claim 2. Including an image generation step of generating the input image based on the ground plane image.
The image generation step is
A step of specifying the trimming position corresponding to the specified tire size based on the data in which the tire size and the trimming position are previously associated with each other.
A step of trimming the ground plane image using the specified trimming position to generate a trimmed image, and
A step of changing the size of the trimmed image to a size that can be input to the feature extraction unit without changing the aspect ratio of the trimmed image to generate the input image.
The method according to claim 1 or 2, wherein the method comprises. A system comprising one or more processors that perform the method according to any one of claims 1-6. A program that causes one or more processors to execute the method according to any one of claims 1 to 6.

Images