Estimation device, estimation method and estimation program, and generation device, generation method and generation program of learned machine learning model

Abstract

To appropriately estimate timing when damage occurs at a tire.SOLUTION: An estimation device includes a derivation part configured to: enter, into one or a plurality of learned machine learning models, input data at least including sensing data on oscillation generated during a rotation of a tire, sensing data on a rotational or acceleration a rotational speed at which a travel test machine rotates the tire, and sensing data on an electric power consumed for rotating the tire by the travel test machine during a durability evaluation of the tire using the travel test machine that rotates the tire at a prescribed rotational speed; and derive output data from the one or plurality of learned machine learning models. The output data corresponds to at least one of: timing when damage occurs at the tire while the tire is rotating; a presence or absence of damage at the tire at a point where the sensing data is acquired; and a portion where the damage occurs at the tire.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an estimation device, an estimation method, and an estimation program for estimating the timing of occurrence of damage to a tire, as well as a generation device, a generation method, and a generation program for a trained machine learning model. [Background technology]

[0002] Patent Document 1 discloses a technique for detecting the occurrence of a failure in a tire component before the tire bursts. The technique disclosed in Patent Document 1 includes a measurement step in which the noise generated when the tire rolls on the road surface is measured by changing the tire's rolling speed stepwise; a calculation step in which a change amount or rate of change in the measurement result is calculated based on the measurement result obtained in the previous measurement step; and a determination step in which the change amount or rate of change is compared with a predetermined threshold and, if this change amount or rate of change exceeds the threshold, a determination is made that a failure has occurred in a tire component. Patent Document 1 also discloses that, instead of noise, the axial force of the tire's mounting hub can be measured and tire failure can be detected based on the measured axial force. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 4259931 Summary of the Invention [Problem to be solved by the invention]

[0004] As disclosed in Patent Document 1, before a tire bursts, abnormalities may be detected in the sound or changes in the axial force. However, there is more than one type of phenomenon that can be a precursor to a burst, and it has been empirically found that a burst can occur even when no noticeable abnormalities in the noise or axial force are detected. This is thought to be because there are various patterns of partial tire damage that can be the starting point for a burst. For this reason, it remains difficult to properly estimate the timing of a burst and the partial tire damage that can cause a burst, and improvements have been desired.

[0005] The present invention aims to provide a technique for appropriately estimating at least one of the timing at which damage will occur in a tire, the presence or absence of damage at a specific point in time, and the location at which damage will occur. [Means for solving the problem]

[0006] An estimation device according to a first aspect of the present invention is an estimation device that estimates at least one of the timing at which damage will occur in a tire and the location at which the damage will occur during a durability evaluation of the tire using a running test machine that rotates the tire at a predetermined rotational speed, and includes a derivation unit that inputs input data to one or more trained machine learning models, the input data including at least sensing data of vibrations occurring during the rotation of the tire, sensing data of the rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of the power consumed to rotate the tire with the running test machine, and derives output data from the one or more trained machine learning models. The output data corresponds to at least one of the timing at which damage will occur in the tire during rotation of the tire, the presence or absence of damage in the tire at the time the sensing data is acquired, and the location at which the damage will occur in the tire.

[0007] An estimation device according to a second aspect of the present invention is the estimation device according to the first aspect, wherein the input data further includes at least one of sensing data of sound generated due to rotation of the tire, sensing data of strain in a portion of the running test machine that imparts rotation to the tire, and sensing data of a load applied to the tire by the running test machine.

[0008] An estimation device according to a third aspect of the present invention is the estimation device according to the first or second aspect, wherein the derivation unit inputs the input data to a plurality of trained machine learning models, each of the plurality of trained models being associated with a respective one of a plurality of parts of the tire, and the output data derived from each of the plurality of trained models corresponds to the presence or absence of damage to each of the plurality of parts at the time the sensing data is acquired.

[0009] An estimation device according to a fourth aspect of the present invention is the estimation device according to any one of the first aspect to the third aspect, wherein the input data further includes at least one of sensing data of the temperature of the rotating tire, sensing data of the internal pressure of the rotating tire, and sensing data of the strain of the rotating tire.

[0010] A fifth aspect of the present invention provides an estimation method executed by one or more computers for estimating at least one of the timing at which damage will occur in a tire and the location at which the damage will occur during a durability evaluation of the tire using a running test machine that rotates the tire at a predetermined rotational speed, the estimation method comprising: inputting input data including at least sensing data of vibrations occurring during the rotation of the tire, sensing data of the rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire with the running test machine into one or more trained machine learning models; and deriving output data from the one or more trained machine learning models. The output data corresponds to at least one of the timing at which damage will occur in the tire during rotation of the tire, the presence or absence of damage in the tire at the time the sensing data was acquired, and the location at which the damage will occur in the tire.

[0011] A sixth aspect of the present invention is an estimation program for estimating at least one of the timing at which damage will occur in a tire and the location at which the damage will occur during a durability evaluation of the tire using a running test machine that rotates the tire at a predetermined rotational speed, the estimation program causing one or more computers to input input data to one or more trained machine learning models, the input data including at least sensing data of vibrations occurring during the rotation of the tire, sensing data of the rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire with the running test machine, and deriving output data from the one or more trained machine learning models. The output data corresponds to at least one of the timing at which damage will occur in the tire during rotation of the tire, the presence or absence of damage in the tire at the time the sensing data was acquired, and the location at which the damage will occur in the tire.

[0012] A seventh aspect of the present invention provides an apparatus for generating a trained machine learning model, the apparatus comprising: an acquisition unit that acquires a training dataset including a plurality of first data acquired during a durability evaluation of a tire using a running test machine that rotates the tire at a predetermined rotational speed, the first data including at least sensing data of vibrations occurring during the rotation of the tire, sensing data of the rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire with the running test machine; and second data that has been combined with the first data in advance; and a learning unit that inputs the first data to a machine learning model, derives output data from the machine learning model, and adjusts parameters that define the machine learning model so as to reduce an error between the output data and the second data. The second data includes at least one of the timing at which damage occurred to the tire during rotation of the tire, whether or not the tire was damaged at the time the sensing data was acquired, and a location at which the damage occurred in the tire.

[0013] A method for generating a trained machine learning model according to an eighth aspect of the present invention is a method for generating a trained machine learning model executed by one or more computers, the method comprising: acquiring a training dataset including a plurality of first data acquired during a durability evaluation of a tire using a running test machine that rotates the tire at a predetermined rotational speed, the first data including at least sensing data of vibrations generated during the rotation of the tire, sensing data of the rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire with the running test machine; and second data pre-combined with the first data; inputting the first data into a machine learning model to derive output data from the machine learning model; and adjusting parameters defining the machine learning model to reduce an error between the output data and the second data. The second data includes at least one of the timing at which damage occurred to the tire during rotation, whether or not the tire was damaged at the time the sensing data was acquired, and the location of the damage in the tire.

[0014] A ninth aspect of the present invention provides a program for generating a trained machine learning model that causes one or more computers to execute the following steps: acquire training datasets including a plurality of first data acquired during a durability evaluation of a tire using a running test machine that rotates the tire at a predetermined rotational speed, the first data including at least sensing data of vibrations occurring during the rotation of the tire, sensing data of the rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire with the running test machine; and second data that has been pre-combined with the first data; input the first data into a machine learning model to derive output data from the machine learning model; and adjust parameters that define the machine learning model so as to reduce an error between the output data and the second data. The second data includes at least one of the timing at which damage occurred to the tire during rotation, whether or not the tire was damaged at the time the sensing data was acquired, and the location of the damage in the tire. [Effects of the Invention]

[0015] According to the present invention, it is possible to use a trained machine learning model to appropriately estimate at least one of the timing at which tire damage will occur, the presence or absence of damage at a specific time point, and the location at which the damage will occur during a tire durability evaluation, thereby enabling more detailed analysis of tire damage and obtaining more insights. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is a diagram showing the overall configuration of a durability test system according to an embodiment; [Figure 2] FIG. 2 is a block diagram showing the electrical configuration of the estimation device according to an embodiment. [Figure 3] FIG. 1 is a diagram illustrating an example of the configuration of a machine learning model. [Figure 4] 1 is a flowchart showing the flow of an estimation method according to an embodiment. [Figure 5]FIG. 2 is a block diagram showing the electrical configuration of a generating device according to an embodiment. [Figure 6] 1 is a flowchart showing the flow of a generation method according to an embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0017] Hereinafter, an estimation device, an estimation method, and an estimation program, as well as a generation device, a generation method, and a generation program for a trained machine learning model according to embodiments of the present invention will be described.

[0018] <1. Durability test system> FIG. 1 is a diagram showing the overall configuration of a tire durability testing system 1 (hereinafter also simply referred to as "system 1") according to one embodiment of the present invention. The system 1 includes an estimation device 2, a running test machine 3, and sensors 4 to 12. The running test machine 3 is a device that rotates a tire T that is a target for durability evaluation. The sensors 4 to 12 are respectively arranged on the running test machine 3, around the running test machine 3, or inside the tire T, and detect physical quantities described below. The estimation device 2 is configured to estimate the timing at which damage will occur to the tire T during a durability evaluation of the tire T using the running test machine 3, based on sensing data output from the sensors 4 to 12. A typical example of damage to the tire T is a burst. Examples of damage other than a burst include partial damage that occurs in various parts such as the tread portion, side portions (e.g., shoulder portions and sidewall portions), and bead portions. Although such partial damage is often not visible in the appearance of the tire T, it can be the cause of a burst, and continued rotation of the tire T is likely to eventually lead to a burst. The estimation device 2 of this embodiment is configured to estimate the timing at which a burst occurs in the tire T and whether or not each portion is damaged at a certain point in time.

[0019] The system 1 coordinates the operations of the estimation device 2, the running test machine 3, and the sensors 4 to 12 using a controller (not shown). For example, the system 1 starts the operations of the estimation device 2, the running test machine 3, and the sensors 4 to 12 in response to a trigger that starts a durability evaluation of the tire T. Furthermore, for example, during a durability evaluation of the tire T, the system 1 stops the running test machine 3 before damage occurs to the tire T or at an early stage of damage based on the estimation result by the estimation device 2. This makes it possible to analyze the state of the tire T before damage or in the early stage of damage, and the knowledge gained can be applied to tire development. Furthermore, events such as a sudden burst of the tire T or continued rotational movement of the tire T despite damage can impose a load on the running test machine 3 and may cause the running test machine 3 to malfunction. According to the system 1, the running test machine 3 is stopped in a timely manner, thereby reducing the load on the running test machine 3.

[0020] <2. Running test machine> The running test machine 3 includes a housing 30, a drive motor 31, a rotating drum 32, a mounting unit 33, and a controller 34. The rotating shaft of the rotating drum 32 is fixed to the housing 30. The rotating drum 32 is driven to rotate at a predetermined speed by the drive motor 31. The mounting unit 33 is also fixed to the housing 30 so as to be movable in the vertical direction. The mounting unit 33 has a mounting hub to which a tire T is rotatably mounted. The tire T is mounted on a predetermined rim and attached to the mounting hub with its internal pressure adjusted to a predetermined value. The vertical position of the mounting unit 33 is controlled by the controller 34 so that the attached tire T is pressed against the outer peripheral surface of the rotating drum 32 with a predetermined load. As a result, when the tire T is rotating normally, the rotational speed and rotational acceleration of the rotating drum 32 match the rotational speed and rotational acceleration of the tire T.

[0021] The controller 34 includes, for example, a CPU (Central Processing Unit), a PLC (Programmable Logic Controller), a ROM (Read Only Memory), and a RAM (Random Access Memory). For example, the ROM stores programs that control the operation of the CPU and the PLC. In addition to controlling the position of the mounting portion 33, the controller 34 also controls the rotation speed of the drive motor 31 and the duration for which rotation continues at a predetermined rotation speed. Multiple combinations of specific rotation speeds of the drive motor 31 and the durations for the specific rotation speeds are stored in advance in a memory such as a ROM. In durability evaluation of the tire T, the controller 34 usually controls the rotation operation of the drive motor 31 according to the pre-stored combinations of rotation speed and duration of the drive motor 31 so that each combination is realized consecutively.

[0022] 3. Sensors The sensors 4 to 12 each detect a predetermined physical quantity during the durability evaluation of the tire T and output time-series sensing data of the detected physical quantity. The time-series sensing data output by the sensors 4 to 12 is sequentially transmitted to the estimation device 2 via a communication line 13 or wirelessly.

[0023] [Power detection sensor] The sensor 4 detects the power consumed by the drive motor 31 to rotate the tire T at a predetermined speed. The sensor 4 is not particularly limited as long as it can detect the power. The sensor 4 detects the power, for example, by acquiring the voltage applied to the drive motor 31 and the power supplied to the drive motor 31. Therefore, the location where the sensor 4 is attached is not particularly limited as long as it is a location where the voltage and current can be acquired, and can be selected appropriately.

[0024] [Rotational speed sensor] The sensor 5 detects the rotational speed at which the rotating drum 32 rotates the tire T (i.e., the rotational speed of the rotating drum 32). The type of sensor 5 is not particularly limited as long as it can detect the rotational speed of a rotating body. For example, a sensor of a type that measures the rotational speed from the output signal of an electromagnetic pickup can be used, or a sensor of a type that generates electricity using rotation, like a dynamo, and measures the rotational speed from the resulting voltage can be used. The mounting location of the sensor 5 is also not particularly limited. For example, the sensor 5 can be mounted on the output shaft of the drive motor 31 or its periphery. The combination of the sensing data output by the sensor 4 and the sensing data output by the sensor 5 indicates the energy loss occurring in the tire T. If damage occurs to the tire T, appropriate friction between the tire T and the rotating drum 32 will not occur, and the above energy loss may be larger or smaller than usual. For this reason, the sensing data that indicates the above energy loss is considered to be correlated with the timing of occurrence of damage to the tire T.

[0025] [Strain sensor] The sensor 6 detects strain at a portion of the running test machine 3 that imparts rotation to the tire T. The sensor 6 is not particularly limited as long as it can detect strain on an object, and known strain gauges, such as those that utilize changes in the electrical resistance of metals, can be used. In this embodiment, the sensor 6 is attached to the inner circumferential surface of the rotating drum 32. However, the location where the sensor 6 is attached is not limited thereto, and the sensor 6 may be attached to the outer circumferential surface of the rotating drum 32 (a location that does not affect the rotation of the tire T) or the output shaft of the drive motor 31. When damage occurs to the tire T, the rotation axis of the tire T becomes unsteady, which may appear as a special strain in the rotating drum 32. For this reason, the strain sensing data is considered to be correlated with the timing at which damage occurs to the tire T. Alternatively, instead of the sensor 6, an optical 3D measurement system including a 3D scanner can be used to measure strain on the tire surface, thereby acquiring strain sensing data.

[0026] [Sound sensor] The sensor 7 detects sound generated due to the rotation of the tire T. The sensor 7 is not particularly limited as long as it can convert sound pressure into an electrical signal, and a known microphone or the like can be used. The sound generated due to the rotation of the tire T may be mainly the contact sound between the tire T and the rotating drum 32, the rotation sound of the drive motor 31 or the rotating drum 32 generated mainly during the rotation of the tire T, or the air sound inside the tire T. In this embodiment, the sensor 7 is attached near the point where the outer circumferential surface of the rotating drum 32 contacts the tire T. However, the attachment location of the sensor 7 is not limited thereto and may be inside the tire T or may be appropriately selected within the room where the running test machine 3 is installed. When damage occurs to the tire T, the sensing data of the sound generated due to the rotation of the tire T may exhibit unusual characteristics. For this reason, it is considered that the sensing data of the sound is correlated with the timing of the occurrence of damage to the tire T.

[0027] [Load sensor] The sensor 8 detects the load applied to the tire T (or the mounting hub) by the running test machine 3. The sensor 8 is not particularly limited as long as it can convert the external force applied to the tire T into an electrical signal, and a known wheel six-component force meter, a load meter installed in the running test machine 3, or the like can be used. The sensing data output by the sensor 8 may be only the load in the vertical direction (z-axis direction), or may include at least one of the load in the direction perpendicular to the rotation axis of the tire T (x-axis direction), the load in the direction parallel to the rotation axis (y-axis direction), and torque around the x, y, and z axes. When damage occurs to the tire T, the load applied to the tire T may deviate from the normal range due to the energy loss or wobble of the rotation axis of the tire T. For this reason, the load sensing data is considered to be correlated with the timing of occurrence of damage to the tire T.

[0028] [Vibration sensor] The sensor 9 detects vibrations generated while the tire T is rotating in the running test machine 3. The sensor 9 is not particularly limited as long as it can convert the frequency of the vibration and at least one of the displacement, velocity, and acceleration into an electrical signal. For example, a piezoelectric sensor, an electrodynamic sensor, a servo sensor, an eddy current sensor, a capacitance sensor, an optical sensor, or the like can be used. In this embodiment, the sensor 9 is installed so as to detect vibrations on the floor 90 of the room in which the running test machine 3 is installed. When the rotation axis of the tire T becomes unstable due to damage or before the tire T bursts, vibrations with unusual characteristics may appear in the running test machine 3 that is rotating the tire T. For this reason, it is considered that the vibration sensing data is correlated with the timing of occurrence of damage to the tire T.

[0029] The sensors 10 to 12 are sensors installed inside the tire T (more precisely, in the space defined by the tire T, the rim, and the wheel) and detect physical quantities of the tire T. By taking into consideration the physical quantities of the tire T in addition to the physical quantities of the running test machine 3 and the environment in which the running test machine 3 is installed, the accuracy of estimation by the estimation device 2 is further improved. The installation locations of the sensors 10 to 12 are not particularly limited as long as they are inside the tire T, and can be selected as appropriate.

[0030] [Temperature sensor] The sensor 10 detects the temperature of the tire T. The sensor 10 is not particularly limited as long as it can convert the temperature of at least one of the rubber and non-rubber components that make up the tire T into an electrical signal, and any sensor can be used. Damage to the tire T occurs due to heat generation and accumulation of mechanical strain. For this reason, it is considered that the sensing data of the temperature of a specific part of the tire T is correlated with the timing of occurrence of damage to that part. Note that the sensor 10 may be installed outside the tire T instead of inside the tire T, and configured to detect the temperature of the tire T in a non-contact manner.

[0031] [Pressure sensor] The sensor 11 detects the internal pressure of the tire T. The sensor 11 is not particularly limited as long as it can convert the internal pressure of the tire T into an electrical signal, and any sensor can be used. When damage occurs to the tire T, the sensing data of the internal pressure of the tire T may exhibit characteristics that are different from normal. For this reason, it is considered that the sensing data of the internal pressure of the tire T is correlated with the timing at which damage occurs to the tire T.

[0032] [Strain sensor] The sensor 12 detects strain in the tire T. The sensor 12 is not particularly limited as long as it can convert the strain in the tire T into an electrical signal, and similar to the sensor 6, any sensor can be used. The sensor 12 can be fixed, for example, to at least one of the inside of the tread portion and the inside of the side portion of the tire T. As described above, damage to the tire T occurs due to the accumulation of mechanical strain. For this reason, it is believed that the sensing data of the strain in the tire T is correlated with the timing at which damage to the tire T occurs.

[0033] <4. Estimation device> 2 is a block diagram showing the electrical configuration of the estimation device 2. The estimation device 2 is a general-purpose computer in terms of hardware, and is realized as an information processing terminal such as a desktop personal computer, a laptop personal computer, a tablet, or a smartphone. The estimation device 2 is manufactured by installing a program 232 into the general-purpose computer from a computer-readable storage medium 233 such as a CD-ROM or a USB memory, or via a network. The program 232 causes the estimation device 2 to perform the operations described below.

[0034] The estimation device 2 includes a control unit 20, a display unit 21, an input unit 22, a storage unit 23, and a communication unit 24. These units 20 to 24 are connected to one another via a bus line and are capable of communicating with one another. The display unit 21 can be configured with a liquid crystal display, an organic EL display, a plasma display, a touch panel display, or the like, and displays output results derived from a machine learning model (described later), etc. The input unit 22 can be configured with a mouse, a keyboard, a touch panel, or the like, and accepts operations on the estimation device 2. The display unit 21 and the input unit 22 may both be configured with the same touch panel display.

[0035] The storage unit 23 can be configured with a non-volatile memory such as a hard disk or a flash memory. In addition to storing a program 232, the storage unit 23 also stores parameters that define trained machine learning models 231A to 231D (hereinafter simply referred to as "trained models 231A to 231D") constructed by machine learning. The trained models 231A to 231D will be described later.

[0036] The control unit 20 can be configured with a CPU, a GPU (Graphics Processing Unit), a ROM, a RAM, etc. The control unit 20 reads and executes a program 232 in the storage unit 23, thereby virtually operating as an acquisition unit 20A, a derivation unit 20B, and a screen generation unit 20C. The acquisition unit 20A acquires measurement values of a tire to be evaluated via the input unit 22, the communication unit 24, etc. The derivation unit 20B inputs input data (described later) to trained models 231A to 231D and derives output data from the trained models 231A to 231D. The screen generation unit 20C generates a screen displaying the (estimated) timing and location of damage occurrence in the tire T, based on the output derived by the derivation unit 20B. The communication unit 24 functions as a communication interface for data communication via a network and data communication with an external device.

[0037] <5. Configuring the trained model> The configurations of the trained models 231A to 231D will be described below. The trained models 231A to 231D are defined by different parameters, and in this embodiment, the input data is common but the output data is different. More specifically, the input data to the trained models 231A to 231D is sensing data output from the sensors 4 to 12 while the tire T is rotating. The time-series sensing data output from the sensors 4 to 12 is treated by the acquisition unit 20A of the estimation device 2 as data sets of the same time that are synchronized with each other, for example, within a predetermined sampling time. When synchronizing the sensing data output from the sensors 4 to 12, the acquisition unit 20A may perform interpolation of any of the sensing data as necessary. The input data is a data set of sensing data synchronized in this way.

[0038] The output data output from the trained model 231A corresponds to the timing at which the tire T begins to burst. This output data can be a value corresponding to the number of seconds from when the running test machine 3 starts rotating the tire until when the tire T begins to burst, for example.

[0039] The output data output from trained models 231B, 231C, and 231D correspond to the presence or absence of damage to the tread, side, and bead portions of tire T, respectively, at the time the sensing data is acquired. That is, trained models 231B, 231C, and 231D are each constructed to estimate the presence or absence of damage to a specific portion of the tire at a certain point in time. These output data can be values corresponding to, for example, the probability that a specific portion has damage (or the probability that a specific portion does not have damage).

[0040] Fig. 3 is a diagram illustrating the configuration of trained models 231A to 231D of this embodiment. As shown in Fig. 3, trained models 231A to 231D are each models based on a neural network (NN). In this embodiment, the trained models 231A to 231D have different parameters (for example, weights w and biases b, which will be described later) that define them, but they share a common layer configuration.

[0041] Each of the machine learning models 231A to 231D includes an input layer 2310 having a predetermined number of nodes, a plurality of intermediate layers 2311, and an output layer 2312. The input layer 2310 is a layer for reading input data, and has the same number of nodes as the number of data included in the input data. Input values are output as is from each node of the input layer 2310, and are input to each node of the first intermediate layer 2311, which is connected immediately after the input layer 2310.

[0042] Each node in the first hidden layer 2311 calculates a value by multiplying the input by a weight w and adding a bias b. The calculated value is then converted using an activation function and output to the node in the subsequent hidden layer 2311. The input here is a plurality of values read by the input layer 2310, and it is possible to appropriately set which node's value in the input layer 2310 is input. The weight w has been adjusted for each value passed from each node in the input layer 2310 to each node in the first hidden layer 2311 through a learning process described below. In other words, for each node in the first hidden layer 2311, weights w corresponding to the number of values input thereto have been optimized. In addition, biases b corresponding to the number of nodes have also been optimized through a learning process described below.

[0043] The activation function is a function for realizing nonlinear transformation. There are no particular limitations on the activation function, and any known activation function such as a tanh function, a sigmoid function, a ReLU function, a step function, an ELU function, or a Softmax function can be used.

[0044] The values output from each node of the first hidden layer 2311 are input to each node of the subsequent hidden layer 2311. The subsequent hidden layer 2311 also has a predetermined number of nodes, and for each node, the weight w and bias b are optimized by the learning process described below. Like the first hidden layer 2311, each node outputs a value calculated for the input value using the corresponding weight w and bias b and an activation function. That is, in principle, each hidden layer 2311 repeats the same processing as the first hidden layer 2311. However, in the last hidden layer 2311, a linear transformation is performed on the input using the optimized weight w and bias b, but a nonlinear transformation using an activation function is not performed. That is, the linearly transformed values output from each node of the last and subsequent hidden layer 2311 are input to each node of the output layer 2312. The number of hidden layers 2311 is preferably two or more, but may be one. It is not particularly limited and can be set appropriately.

[0045] The output layer 2312 has nodes whose number corresponds to the output data, and each node typically outputs the input value as is. In this embodiment, the output layer 2312 has one node. The data output from the output layer 2312 becomes the output data of each of the trained models 231A to 231D. The output from the output layer 2312 of the trained model 231A is a value corresponding to the number of seconds from when the tire starts rotating to when the tire T starts bursting. The output from the output layer 2312 of the trained model 231B is a value corresponding to the probability that there is damage to the tread portion of the tire T at the time when the sensing data that serves as input data is acquired. The output from the output layer 2312 of the trained model 231C is a value corresponding to the probability that there is damage to the side portion of the tire T at the time when the sensing data is acquired. The output from the output layer 2312 of the trained model 231D is a value corresponding to the probability that there is damage to the bead portion of the tire T at the time when the sensing data is acquired.

[0046] <6. Estimation method> 4 is a flowchart showing the flow of the estimation method executed by the estimation device 2 according to this embodiment. The following estimation method starts, for example, when the tire T is set on the running test machine 3 and starts to rotate, and is repeated almost in real time until the running test machine 3 is stopped, and ends when the running test machine 3 is stopped.

[0047] First, the acquiring unit 20A sequentially acquires sensing data outputted from the sensors 4 to 12 from time to time (step S1). The acquiring unit 20A temporarily stores the acquired sensing data in the RAM or stores the data in the storage unit .

[0048] Next, the acquisition unit 20A synchronizes the time-series sensing data from each sensor stored in step S1 based on a predetermined algorithm to generate a data set of each sensing data for each time (step S2). The acquisition unit 20A temporarily stores the generated data set in RAM or stores it in the storage unit 23.

[0049] Next, the derivation unit 20B inputs the data set generated in step S2 into the trained models 231A to 231D and derives each output data corresponding to this input data from the trained models 231A to 231D (step S3). Here, if it is estimated based on each output data that the tire T will soon burst or that damage has occurred in any part of the tire T, the derivation unit 20B may be configured to notify the controller of the system 1 of this. The threshold value of the output data estimated as described above may be set appropriately. Furthermore, for a plurality of chronologically consecutive output data derived corresponding to a plurality of chronologically consecutive input data, the derivation unit 20B may be configured to notify the controller of the system 1 of this when, for example, the number of seconds until the tire T begins to burst is gradually decreasing or the probability that damage has occurred in any part of the tire T is gradually increasing. Upon receiving this notification, the system 1 can control the running test machine 3 to immediately stop the running test machine 3.

[0050] Next, the screen generation unit 20C generates a result screen that displays the results based on the output data derived in step S3, and displays it on the display unit 21 (step S4). The result screen may be a screen that shows the values of the output data from the trained models 231A to 231D, or, in addition to or instead of this, may be a screen that shows the remaining time until the tire T is estimated to burst or the part of the tire T that is estimated to be damaged. By checking the result screen, the user of the system 1 can recognize that the tire T is in a state where it may burst, or that there is a high possibility that damage has occurred in a specific part of the tire T.

[0051] Next, the acquisition unit 20A determines whether the running test machine 3 is currently stopped (step S5). When the acquisition unit 20A determines that the running test machine 3 is not stopped (NO), it executes the processes from step S1 onwards again. That is, based on the latest sensing data, it is estimated whether the tire T is in a state where it may burst and whether any part of the tire T is damaged.

[0052] On the other hand, when the acquisition unit 20A determines that the running test machine 3 has stopped (YES), the estimation process by the estimation device 2 ends.

[0053] According to the above steps S1 to S5, although there is a slight time lag strictly speaking with respect to the time at which the sensing data of the sensors 4 to 12 are output, the running test machine 3 can be stopped before a burst occurs in the tire T or at a time around the time when partial damage occurs in the tire T. This makes it possible to discover damage to the tire T at a relatively early stage and suppress damage to the running test machine 3 due to a burst of the tire T. Furthermore, it is possible to analyze the cause of a burst in the tire T and the progression of partial damage.

[0054] If, due to the calculation speed of the estimation device 2, a large time lag occurs between the acquisition of sensing data by the sensors 4 to 12 and the derivation of an estimation result based on this sensing data, the following can also be done. For example, based on past data from similar tires, the tire T is rotated by the running test machine 3 until a time when it is thought that a burst will not occur, and time-series sensing data from the sensors 4 to 12 is acquired. After the running test machine 3 is stopped, input data generated based on the sensing data is input to the trained models 231A to 231D, and the timing when the tire T will burst, etc., is estimated.

[0055] <7.Generation device> FIG. 5 is a block diagram showing the electrical configuration of a generating device 2X configured to execute a method for generating trained models 231A to 231D. The generating device 2X is a general-purpose computer in terms of hardware, and is realized as an information processing terminal such as a desktop personal computer, a laptop personal computer, a tablet, or a smartphone. The generating device 2X is manufactured by installing a program 232X into a general-purpose computer from a computer-readable storage medium 233X such as a CD-ROM or a USB memory, or via a network. The program 232X causes the generating device 2X to perform the operations described below.

[0056] The generation device 2X includes a control unit 20X, a display unit 21X, an input unit 22X, a storage unit 23X, and a communication unit 24X. These units 20X to 24X are connected to one another via a bus line and can communicate with each other. The same explanations as for the display unit 21, input unit 22, and communication unit 24 of the estimation device 2 apply to the display unit 21X, input unit 22, and communication unit 24, and therefore further description will be omitted.

[0057] The control unit 20X can be configured with a CPU, a GPU (Graphics Processing Unit), a ROM, a RAM, etc. The control unit 20X reads and executes a program 232X in the storage unit 23X, thereby virtually operating as an acquisition unit 20D and a learning unit 20E. The acquisition unit 20D acquires training data sets 234A to 234D from outside the generation device 2X via a storage medium 233X, a communication unit 24X, etc. The training data sets 234A to 234D are data sets for adjusting parameters of the trained models 231A to 231D, respectively.

[0058] The storage unit 23X can be configured with a nonvolatile memory such as a hard disk and a flash memory, etc. In the storage unit 23X, a program 232X is stored, and also learning data sets 234A to 234D are stored.

[0059] The training data sets 234A to 234D are data sets created from data on durability evaluations of multiple tires using the running test machine 3. The multiple tires preferably have a common size and the same or similar tread pattern. Furthermore, the durability evaluation conditions (rotational speed and its duration) are assumed to be common to the multiple tires. Each of the training data sets 234A to 234D includes multiple first data and multiple second data to be combined with the first data. The first data is common to the training data sets 234A to 234D and corresponds to the input data of the trained models 231A to 231D. The first data is created from time-series sensing data output from the sensors 4 to 12 during a predetermined time range from 0 to t seconds during the durability evaluation of one tire. 0 second is the start time by the running test machine 3, and t seconds represents the specified duration of the durability evaluation (the length of time the tire rotates). The time-series sensing data is a collection of consecutive data sets synchronized over a time range that can be considered to be at the same time. The first data is a collection of the above data sets for one tire over a time range from 0 to t seconds, and the number of data sets is predetermined. However, for a tire that bursts during durability testing, the data set from the time t0 seconds when the burst began onwards is filled with a predetermined value. The multiple first data are collections of first data for multiple tires.

[0060] The second data is data to be combined with the first data and has the same number of values as the first data. The second data of the training dataset 234A is the timing (number of seconds t0) at which the tire of the corresponding first data started to burst after the tire started to rotate. If the tire did not burst, the second data is number of seconds t.

[0061] The second data in the training dataset 234B is the probability that damage has occurred in the tread portion of the tire of the corresponding first data. The second data is, for example, a value ranging from 0 (no damage to the tread portion) to 1 (damage to the tread portion). In other words, all of the second data for a tire for which no damage to the tread portion was confirmed after the durability evaluation is 0. On the other hand, at least some of the second data for a tire for which damage to the tread portion was confirmed after the durability evaluation is 1. The second data for a tire for which damage to the tread portion was confirmed can be created, for example, as follows: First, all of the second data to be combined with the first data in a first time range in which it is estimated that no damage has occurred in the tread portion is set to 0. Next, all of the second data to be combined with the first data in a second time range in which it is estimated that damage has definitely occurred in the tread portion is set to 1. If the first time range and the second time range are not consecutive, all of the second data to be combined with the first data in the time range between them is set to 0.5.

[0062] The second data of training data set 234C is the probability that damage has occurred in the side portion of the tire of the corresponding first data. The second data of training data set 234D is the probability that damage has occurred in the bead portion of the tire of the corresponding first data. The second data of training data sets 234C and 234D can each be created in the same manner as the second data of training data set 234B.

[0063] <8. Generation method> FIG. 6 is a flowchart showing the flow of a method for generating trained models 231A to 231D, which is executed by a generation device 2X according to this embodiment.

[0064] First, the acquisition unit 20D acquires the training data sets 234A to 234D from outside the generation device 2X via the storage medium 233X, the communication unit 24X, etc. (Step S11). The acquisition unit 20D stores the acquired training data sets 234A to 234D in the storage unit 23X.

[0065] Next, the learning unit 20E divides each of the learning data sets 234A-234D into data sets for a predetermined number of tires, and creates multiple subsets each including data sets for the same number of tires (step S12). The predetermined number is the number of data to be continuously input to each machine learning model in the next step S13, and can be set as appropriate. By repeating steps S12-S16, each machine learning model becomes a trained model 231A-231D. In other words, each machine learning model has the layer structure shown in FIG. 3, and is a model in a state where parameters have not been adjusted (i.e., before learning).

[0066] Next, the learning unit 20E selects one of the subsets, sequentially inputs the first data included in the selected subset to each machine learning model, and derives output data from each machine learning model corresponding to the input first data (step S13). The output data corresponds to the second data combined with the first data input as input data.

[0067] Next, the learning unit 20E adjusts and updates parameters so that the value of the error function between the output data derived in step S13 and the second data combined with the first data input in step S13 becomes minimum (step S14). More specifically, the learning unit 20E adjusts and updates weight coefficients, biases, etc. in the intermediate layer of the machine learning model by gradient descent or the like.

[0068] Next, the learning unit 20E determines whether one epoch of learning has been completed (step S15). In this embodiment, it is determined that one epoch of learning has been completed when steps S13 and S14 have been performed once for each subset created in step S12. If it is determined that one epoch of learning has not been completed (NO), the learning unit 20E repeats steps S13 to S14 using a subset that has not yet been used. On the other hand, if it is determined that one epoch of learning has been completed (YES), step S16 is executed. Note that in this process, a dropout method in which nodes are randomly eliminated may be applied to prevent overlearning.

[0069] In the following step S16, the learning unit 20E determines whether learning of all epochs has been completed. The total number of epochs is not particularly limited and can be set as appropriate. If it is determined that learning of all epochs has not been completed (NO), the learning unit 20E repeats steps S12 to S14 while again generating subsets of different combinations. On the other hand, if it is determined that learning of all epochs has been completed (YES), the learning unit 20E stores the latest parameters in the storage unit 23X and uses these as parameters that define the trained models 231A to 231D. In other words, trained models 231A to 231D are generated by the above procedure.

[0070] Note that when acquiring the learning data sets 234A-234D, the acquiring unit 20D may separate each of these data sets into training data and test data and store them in the storage unit 23X. Then, while executing steps S12-S15 for the training data, the learning unit 20E may input the first data of the test data to the current machine learning model at an appropriate time, calculate the error between the output data and the second data of the test data, and display the calculation result on the display unit 21X. In this way, if it is considered that the output error of the machine learning model has converged within a predetermined range before the completion of learning for all epochs, the learning may be terminated at that point.

[0071] <9. Variations> Although one embodiment of the present invention has been described above, the present invention is not limited to the above embodiment, and various modifications are possible without departing from the spirit of the present invention. The gist of the following modifications can be combined as appropriate.

[0072] (1) In the above embodiment, NN-based models were used as the trained models 231A to 231D. However, the machine learning models are not limited to this. Other machine learning models, such as models using support vector machines (SVMs), convolutional neural network (CNN) models, K-NN models, clustering, k-means, decision trees, and logistic regression models, and models combining these may also be used. Furthermore, even in the case of NN-based models, the layer structure and the like may be changed as appropriate. The method of generating the trained models 231A to 231D is not particularly limited to the above embodiment, and known methods may be applied.

[0073] (2) In the above embodiment, the input data for the trained models 231A to 231D was one data set of synchronized sensing data. However, the input data may be a predetermined number of data sets of synchronized sensing data that are consecutive in time series. It is believed that using a continuous time series data set as input data within a range that does not affect the speed estimated by the estimation device 2 makes it easier to detect damage to the tire T.

[0074] (3) In the above embodiment, the output data of trained model 231A corresponded to the time until tire T began to burst. However, the output data of trained model 231A may correspond to the probability of burst occurrence at the time when the sensing data input as input data was acquired, similar to the output data of trained models 231B to 231D in the above embodiment. In this case, the second data of training dataset 234A can be created in the same manner as the second data of training dataset 234B in the above embodiment. Furthermore, the time until tire T begins to burst may not be based on the start of rotation of tire T by running test machine 3, but may be based on a predetermined point in time after tire T begins to rotate.

[0075] (4) Any of the trained models 231A to 231D of the estimation device 2 may be omitted. That is, the estimation device 2 may be configured to estimate any of the following: the timing at which a tire begins to burst; the presence or absence of damage to the tread portion at the time of acquiring the sensing data; the presence or absence of damage to the side portion at the time of acquiring the sensing data; or the presence or absence of damage to the bead portion at the time of acquiring the sensing data. Furthermore, the trained models 231B to 231D may be configured as a single integrated trained model. In this case, for example, three nodes corresponding to the tread portion, the side portion, and the bead portion may be created in the output layer of one trained model, and each node may output a value corresponding to the probability of damage occurring at the time of acquiring the sensing data. That is, the output data in this case corresponds to both the presence or absence of damage to the tire T at the time of acquiring the sensing data and the location of the damage occurring on the tire T.

[0076] (5) The input data to trained models 231A to 231D need only include sensing data of vibrations occurring during rotation of tire T, sensing data of the rotational acceleration or rotational speed at which tire T is rotated by running test machine 3, and sensing data of the power consumed to rotate tire T by running test machine 3. In other words, at least one of sensors 6 to 8 and sensors 11 to 12 may be omitted.

[0077] (6) In the above embodiment, the estimation device 2 and the generation device 2X are each configured as a single device, but they may also be configured as a single device. Alternatively, the functions of the units 20A to 20C, 20D, and 20E, the storage unit 23, and the storage unit 23X may be distributed across multiple devices. For example, learning of the machine learning model may be performed using a service provided over a network, and the constructed trained models 231A to 231D may be stored in the estimation device 2. Furthermore, the trained models 231A to 231D may be constructed in a distributed manner across multiple devices.

[0078] (7) The control units 20 and 20X may be configured to include a CPU, a GPU, a vector processor, an FPGA (Field Programmable Gate Array), an ASIC (Application Specific Integrated Circuit), or other chips dedicated to artificial intelligence. The operations of the control units 20 and 20X may be executed by one or more processors. Furthermore, at least one of the control units 20, 20X, and the controller 34 may also serve as the controller of the system 1.

[0079] 1 System 2 Estimation device 2X generator 3. Running test machine 4~12 sensors 231A~231D Trained Models

Claims

1. An estimation device that estimates at least one of a timing when damage occurs in a tire and a location where the damage occurs during a durability evaluation of the tire using a running test machine that rotates the tire at a predetermined rotation speed, a derivation unit that inputs input data including at least sensing data of vibrations occurring during rotation of the tire, sensing data of rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire by the running test machine into one or more trained machine learning models, and derives output data from the one or more trained machine learning models. Equipped with The output data corresponds to at least one of a timing at which damage occurs to the tire during rotation of the tire, whether or not damage exists in the tire at the time the sensing data is acquired, and a location at which damage occurs in the tire. Estimation device.

2. The input data further includes at least one of sensing data of a sound generated due to rotation of the tire, sensing data of strain in a portion of the running test machine that imparts rotation to the tire, and sensing data of a load applied to the tire by the running test machine. The estimation device according to claim 1 .

3. The derivation unit inputs the input data into a plurality of trained machine learning models; Each of the plurality of trained models is associated with each of a plurality of parts of the tire; The output data derived from each of the plurality of trained models corresponds to the presence or absence of damage to each of the plurality of parts at the time of acquiring the sensing data. The estimation device according to claim 1 or 2.

4. the input data further includes at least one of sensing data of a temperature of the rotating tire, sensing data of an internal pressure of the rotating tire, and sensing data of a strain of the rotating tire. The estimation device according to claim 1 or 2.

5. An estimation method executed by one or more computers for estimating at least one of timing when damage will occur in a tire and a location where the damage will occur during a tire durability evaluation using a running test machine that rotates the tire at a predetermined rotational speed, the method comprising: inputting input data including at least sensing data of vibrations occurring during rotation of the tire, sensing data of rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire by the running test machine into one or more trained machine learning models, and deriving output data from the one or more trained machine learning models; Including, The output data corresponds to at least one of a timing at which damage occurs to the tire during rotation of the tire, whether or not damage exists in the tire at the time the sensing data is acquired, and a location at which damage occurs in the tire. Estimation method.

6. An estimation program for estimating at least one of timing and location of damage to a tire during a durability evaluation of the tire using a running test machine that rotates the tire at a predetermined rotation speed, the estimation program comprising: inputting input data including at least sensing data of vibrations occurring during rotation of the tire, sensing data of rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire by the running test machine into one or more trained machine learning models, and deriving output data from the one or more trained machine learning models; Execute The output data corresponds to at least one of a timing at which damage occurs to the tire during rotation of the tire, whether or not damage exists in the tire at the time the sensing data is acquired, and a location at which damage occurs in the tire. Estimation program.

7. an acquisition unit that acquires a learning dataset including a plurality of first data acquired during a tire durability evaluation using a running test machine that rotates the tire at a predetermined rotational speed, the first data including at least sensing data of vibrations occurring during the rotation of the tire, sensing data of rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire with the running test machine, and second data that is combined with the first data in advance; a learning unit that inputs the first data into a machine learning model, derives output data from the machine learning model, and adjusts parameters that define the machine learning model so that an error between the output data and the second data is reduced; Equipped with The second data includes at least one of a timing when damage to the tire occurs during rotation of the tire, whether or not the tire is damaged at the time when the sensing data is acquired, and a location where the damage occurs in the tire. A device for generating trained machine learning models.

8. 1. A method for generating a trained machine learning model, executed by one or more computers, comprising: Acquiring a learning dataset including a plurality of first data acquired during a tire durability evaluation using a running test machine that rotates the tire at a predetermined rotational speed, the first data including at least sensing data of vibrations occurring during the rotation of the tire, sensing data of rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire with the running test machine, and second data that is combined with the first data in advance; inputting the first data into a machine learning model and deriving output data from the machine learning model; adjusting parameters that define the machine learning model so that an error between the output data and the second data is reduced; Including, The second data includes at least one of a timing when damage to the tire occurs during rotation of the tire, whether or not the tire is damaged at the time when the sensing data is acquired, and a location where the damage occurs in the tire. How to generate a trained machine learning model.

9. A program for generating a trained machine learning model, which is installed on one or more computers: Acquiring a learning dataset including a plurality of first data acquired during a tire durability evaluation using a running test machine that rotates the tire at a predetermined rotational speed, the first data including at least sensing data of vibrations occurring during the rotation of the tire, sensing data of rotational acceleration or rotational speed at which the running test machine rotates the tire, and sensing data of power consumed to rotate the tire with the running test machine, and second data that is combined with the first data in advance; inputting the first data into a machine learning model and deriving output data from the machine learning model; adjusting parameters that define the machine learning model so that an error between the output data and the second data is reduced; Execute The second data includes at least one of a timing when damage to the tire occurs during rotation of the tire, whether or not the tire is damaged at the time when the sensing data is acquired, and a location where the damage occurs in the tire. A program for generating trained machine learning models.

Images