Method and system for inspecting splice portions of tire components, and method for manufacturing tires.
The method and system use machine learning to assess splice portions in tire components, ensuring uniformity and reducing defects by predicting RFV data fluctuations, thereby enhancing tire manufacturing efficiency.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- THE YOKOHAMA RUBBER CO LTD
- Filing Date
- 2024-12-09
- Publication Date
- 2026-06-19
Smart Images
Figure 2026100235000001_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a method and a system for inspecting a splice portion of a tire member, and a method for manufacturing a tire. More specifically, the present invention relates to a method and a system for inspecting a splice portion of a tire member, which can accurately and efficiently detect whether a tire member is spliced without impairing good uniformity of the tire, and a method for manufacturing a tire using this inspection method.
Background Art
[0002] In the tire manufacturing process, one end and the other end of a longitudinal direction of a tire member made of unvulcanized rubber such as a cap tread are spliced on a forming drum and formed into a cylindrical shape. If the splice amount of the tire member is too small, the bonding strength may be insufficient and the splice portion may peel off. If the splice portion peels off and a gap is formed, the periphery of the gap becomes thinner than the non-splice portion. If the splice amount is too large, the thickness of the splice portion becomes larger than that of the non-splice portion, and the difference becomes more significant. Since the excess or deficiency of the splice amount greatly affects the uniformity of the tire, it is important to ensure an appropriate splice amount.
[0003] Conventionally, a method for detecting an abnormal joint of a splice portion has been proposed (see Patent Document 1). In the method proposed in Patent Document 1, the shape of the splice portion of a tire component joined on a rotating drum is captured by an image input processing means, and the presence or absence of an abnormality in the splice portion is determined based on the captured data. In this method, the presence or absence of an abnormality in the splice portion is determined using, as indices, the splice amount, the parallelism and straightness of the boundary line of the splice portion, etc. (paragraph 0015, etc.). In this method, there is no clear link between the presence or absence of an abnormality in the splice portion and the degree of uniformity of the tire. Therefore, there is room for improvement in accurately and efficiently detecting whether a tire member is spliced without impairing good uniformity of the tire.
Prior Art Documents
Patent Documents
[0004] [Patent Document 1] Japanese Patent Application Publication No. 6-166121 [Overview of the project] [Problems that the invention aims to solve]
[0005] The object of the present invention is to provide a method and system for inspecting splice portions of tire components that can accurately and efficiently detect whether tire components are spliced without impairing the good uniformity of the tire, and a method for manufacturing a tire using this inspection method. [Means for solving the problem]
[0006] The present invention provides a method for inspecting the splice portion of a tire component, characterized in that a predictive model is constructed and stored in a computing device, which determines the suitability of the splice portion based on the magnitude of the fluctuation of the RFV data, by machine learning using a dataset of image data showing the state of the splice portion of a predetermined tire component acquired in the molding process of a green tire, and fluctuation data showing the degree of fluctuation of RFV data in the region corresponding to the splice portion, acquired by uniformity measurement of a tire manufactured by vulcanizing the green tire, as training data, and the image data showing the state of the splice portion of the predetermined tire component of the green tire to be inspected is input into the predictive model, thereby determining the suitability of the splice portion of the green tire to be inspected at the stage of the molding process of the green tire to be inspected.
[0007] The present invention provides an inspection system for the splice portion of a tire member, comprising: an image acquisition device that acquires image data showing the state of a predetermined splice portion of a tire member during the molding process of a green tire; and a computing device, wherein the computing device stores a prediction model that is constructed by machine learning using a dataset of the image data and fluctuation data showing the degree of fluctuation of RFV data in the region corresponding to the splice portion acquired by uniformity measurement of a tire manufactured by vulcanizing the green tire as training data, and which determines the suitability of the splice portion based on the magnitude of the fluctuation of the RFV data, and the image data showing the state of the predetermined splice portion of the green tire to be inspected is input to the prediction model, so that the suitability of the splice portion of the green tire to be inspected is determined at the stage of the molding process of the green tire to be inspected.
[0008] The present invention's method for manufacturing a tire is characterized by manufacturing a tire by vulcanizing a green tire that has been determined to have an appropriate splice portion of a predetermined tire member based on the above-described inspection method for the splice portion of the tire member. [Effects of the Invention]
[0009] According to the tire member splice inspection method and inspection system of the present invention, by inputting the image data showing the state of the splice portion of the predetermined tire member of the green tire to be inspected into the prediction model, the magnitude of the fluctuation in the RFV data of the tire manufactured by vulcanizing the green tire to be inspected can be predicted. Then, based on the predicted magnitude of the fluctuation in the RFV data, the suitability of the splice portion of the predetermined tire member of the green tire to be inspected is determined at the molding stage of the green tire to be inspected. As a result, it becomes possible to accurately and efficiently detect whether the predetermined tire member of the green tire to be inspected is spliced without impairing the good uniformity of the tire.
[0010] According to the tire manufacturing method of the present invention, since the green tire to be inspected, in which the splice portion of the predetermined tire member is judged to be appropriate, is vulcanized to produce a tire with excellent uniformity, a tire can be obtained. Furthermore, since the appropriateness of the splice portion of the predetermined tire member is judged to be appropriate at the molding stage of the green tire to be inspected, appropriate measures can be taken early on the green tire to be inspected in which the splice portion is judged to be inappropriate. As a result, it is advantageous for improving tire yield. [Brief explanation of the drawing]
[0011] [Figure 1] This is an explanatory diagram illustrating an example of a tire component inspection system. [Figure 2] Figure 1 is an explanatory diagram illustrating the drum body and tire components in a front view. [Figure 3] This is an explanatory diagram illustrating a vulcanized tire in a cross-sectional view. [Figure 4] Figure 3 is an explanatory diagram illustrating the position of the splice portion of the tire component in a tire, viewed from the side. [Figure 5] This is an explanatory diagram illustrating, in a schematic manner, image data of the spliced and unspliced portions of the cap tread acquired by the image acquisition device shown in Figure 1. [Figure 6] This is an explanatory diagram illustrating RFV data obtained by measuring the uniformity of a tire. [Figure 7] Figure 6 is an explanatory diagram illustrating the degree of variation in the region corresponding to the splice portion of the RFV data in the cap tread. [Modes for carrying out the invention]
[0012] The inspection method and inspection system for tire components and the manufacturing method for tires according to the present invention will be described below based on the embodiments shown in the figures.
[0013] In the molding process of the green tire G illustrated in Figures 1 and 2, a predetermined tire member E is wrapped around the outer surface of a drum body 6, and its longitudinal end and other end are spliced together to form a cylindrical shape, with the spliced portion becoming the splice portion Es. The longitudinal end and other end of the tire member E to be spliced are generally cut at the same angle in the thickness direction, overlapping each other during splicing. The drum body 6 rotates around a central axis 6a. Various known types of tire molding drums and rigid cores can be used as the drum body 6. The rigid core is a drum body 6 having an outer surface that is substantially the same shape as the inner surface of the tire T manufactured by vulcanizing the green tire G.
[0014] The embodiment of the tire component inspection system 1 determines whether the splice portion Es of a predetermined tire component E of the green tire Gt to be inspected is properly spliced at the stage of the molding process of the green tire Gt. The arrows W, R, and C in the figure indicate the width direction, radial direction, and circumferential direction of the green tire G, Gt, and tire T, respectively.
[0015] As illustrated in Figure 3, a tire T manufactured by vulcanizing a green tire G is composed of multiple types of tire components E integrated together. In a typical tire T, an inner liner layer E3 is located at the innermost circumference, and a carcass layer is located on its outer circumference. The carcass layer is mounted between a pair of left and right bead sections, and both ends of the carcass layer are folded back from the inside to the outside of the tire around the bead core of each bead section. A belt layer is located on the outer circumference of the center of the carcass layer in the tire width direction, and a cap tread E1 is located on its outer circumference. Side treads E2 are located on both sides of the cap tread E1 in the tire width direction. Side rubber is located on the outer circumference of the carcass layer between each side tread E2 and each bead section. The tire T is not limited to the structure illustrated in Figure 3, and other tire components E may be appropriately arranged as needed.
[0016] In the tire T manufactured by vulcanizing the green tire G, as illustrated in FIG. 4, there is a splice portion Es of each tire member E. More specifically, with respect to a preset reference position (reference mark M) of the tire T, a splice portion Es1 of the cap tread E1 exists at a position of an angle A1 counterclockwise around the tire axis. Similarly, a splice portion Es2 of the side tread E2 is located at the position of the angle A2, and a splice portion Es3 of the inner liner E3 is located at the position of the angle A3. As the reference mark M, for example, a QR label storing the unique information of the tire T is used. The reference mark M is set and installed in the molding process of the green tires G and Gt.
[0017] As illustrated in FIG. 1, the inspection system 1 includes an image acquisition device 2 and an arithmetic unit 3. The image acquisition device 2 is connected to the arithmetic unit 3 via an input unit 4. An output unit 5 such as a known monitor or indicator light is connected to the arithmetic unit 3 by wire or wirelessly.
[0018] The image acquisition device 2 acquires image data Ds indicating the state of the splice portion Es of a predetermined tire member E in the molding process of the green tires G and Gt. More specifically, as the splice portion Es of the predetermined tire member E, the image acquisition device 2 acquires front view image data Ds with a region including the boundary between the splice portion Es and the non-splice portion Ex as the main range. Therefore, this image data Ds necessarily includes the boundary between the splice portion Es and the non-splice portion Ex appearing on the surface of the tire member E.
[0019] Examples of the predetermined tire member E include at least one of the cap tread E1, the side tread E2, and the inner liner E3. The cap tread E1, the side tread E2, and the inner liner E3 are each formed substantially only of unvulcanized rubber.
[0020] Each image data Ds is acquired under basically the same conditions. As the image acquisition device 2, various known profile sensors, digital camera devices, etc. are adopted. In the image data Ds, the difference in the thickness (shape) of the tire member E is grasped as a difference in shading.
[0021] Since the drum body 6 rotates about the central axis 6a, by rotating the drum body 6 about the central axis 6a to shift the circumferential position of the tire member E, the front view image data Dx of the non-splice portion Ex of the tire member E can be acquired by the image acquisition device 2.
[0022] That is, according to this embodiment, as illustrated in FIG. 5, the front view image data Ds and Dx of the splice portion Es and the non-splice portion Ex of the tire member E can be acquired by the image acquisition device 2. Therefore, it is also possible to acquire the front view image data Ds and Dx over the entire circumferential length of the tire member E. The image data Dx is also acquired under the same conditions as the image data Ds.
[0023] In addition, in FIG. 5, the region of the oblique line extending diagonally downward to the right is schematically shown to be slightly convex, and the region of the oblique line extending diagonally downward to the left is schematically shown to be slightly concave. The smaller the arrangement pitch of the oblique lines, the larger the thickness is schematically shown. Therefore, the tire member E illustrated in FIG. 5 has a plurality of undulations in an uneven shape in the width direction, and each convex and concave shape extends in the circumferential direction. The splice portion Es is slightly thicker than the non-splice portion Ex. Such a state of the tire member E can be grasped from the image data Ds and Dx.
[0024] The arithmetic device 3 can receive the image data Ds and Dx acquired by the image acquisition device 2. As the arithmetic device 3, various known computers are used. The arithmetic device 3 has a storage unit 3a such as a memory, an arithmetic unit 3b such as a CPU, and an input unit 4. The input unit 4 is an interface portion to which various devices are connected to the arithmetic device 3, and various data are input and stored in the storage unit 3a through the input unit 4.
[0025] The arithmetic unit 3 performs various calculations using the input and stored data. The memory unit 3a stores the prediction model PM, which will be described later, and the arithmetic unit 3b performs calculations using the prediction model PM.
[0026] Furthermore, the image data Ds and Dx input to the arithmetic unit 3 will not include any extraneous parts other than the tire components E, such as the drum body 6. In addition, if the image data Ds and Dx contain dust or scratches and irregularities that are clearly likely to cause noise, they will be removed from the image data Ds and Dx by applying known noise processing (filtering).
[0027] Output unit 5 outputs the calculation results (judgment results) using the prediction model PM. If a monitor is used as output unit 5, image data Ds and Dx acquired by the image acquisition device 2, and various calculation processing results from the calculation device 3 can be output.
[0028] The predictive model PM is a type of computer program that determines the suitability of a splice portion Es of a predetermined tire component E of a green tire Gt under inspection. The predictive model PM is constructed by machine learning using a dataset of image data Ds of a predetermined tire component E of a green tire G and fluctuation data Dv, which will be described later, as training data. The fluctuation data Dv shows the degree of fluctuation of RFV data in the region corresponding to the splice portion Es of the predetermined tire component E, which is obtained by measuring the uniformity of a tire T manufactured by vulcanizing the green tire G from which the image data Ds was acquired.
[0029] More specifically, the RFV data of the tire T exemplified in Figure 6 is acquired by various known tire uniformity measuring devices. In Figure 6, the vertical axis represents the measured RFV value, and the horizontal axis represents the phase (circumferential position) of the tire T with the reference mark M as the origin (0°), showing the RFV data for one full rotation of the tire T. Es1, Es2, and Es3 represent the regions (phases) corresponding to the splice portion Es of the cap tread E1, side tread E2, and inner liner E3, respectively. In the molding process of the green tire G, a reference mark M is attached, and the position of the splice portion Es relative to the reference mark M is known, so the region (phase) corresponding to the splice portion Es of a given tire member E can be identified in the RFV data exemplified in Figure 6.
[0030] The magnitude of the difference between the minimum and maximum RFV data, known as PP, serves as an indicator of the uniformity performance of the tire T. A smaller PP value indicates better uniformity. Therefore, to improve the uniformity performance of the tire T, the value of this PP value should be reduced.
[0031] The variation data Dv should indicate the degree of variation in the RFV data in the region corresponding to the splice Es. Examples of variation data Dv include the protrusion height of the RFV data in the region corresponding to the splice Es, or the area of the shaded portion exemplified in Figure 7.
[0032] Figure 7 shows the variation data Dv for the cap tread E1. To calculate this variation data Dv, first, the maximum value phase Pm, which indicates the maximum value of the RFV data in the region corresponding to the splice Es1, is identified. Then, the area of the region (shaded area in Figure 7) that protrudes to the positive side of the RFV data, enclosed by the line segment connecting the intersection points of the RFV data and the nearby front-to-back phases (Pm-α, Pm+α) centered on the maximum value phase Pm, is used as the variation data Dv. The value of α is set as appropriate, but for example, it is about 10° or 15°. Similarly, the area of the shaded area can be calculated and used as variation data Dv for the side tread E2 and inner liner E3.
[0033] The predictive model PM is constructed by using a computer to perform machine learning on numerous datasets of the aforementioned image data Ds and variable data Dv as training data. Examples of machine learning methods include various well-known techniques such as deep learning using neural networks. The computer used to construct the predictive model PM may be the computing unit 3 that constitutes the inspection system 1, or it may be a computer separate from the inspection system 1.
[0034] The prediction model PM receives image data Ds indicating the state of the splice portion Es of a predetermined tire component E of the green tire Gt under inspection. Based on the input image data Ds, it predicts the magnitude of fluctuation data Dv in the region corresponding to the splice portion Es of the tire component E in the tire T manufactured by vulcanizing the green tire Gt. The prediction model PM is then configured to determine whether the splice portion Es of the predetermined tire component E is suitable or unsuitable based on the predicted magnitude of fluctuation data Dv, i.e., the magnitude of the fluctuation in RFV data.
[0035] This section describes an example of the procedure for constructing this predictive model (PM).
[0036] First, when forming the green tire G, image data Ds showing the state of the splice portion Es1 of a predetermined tire component E (in this embodiment, the cap tread E1) is acquired. Then, uniformity measurement is performed on the tire T manufactured by vulcanizing the green tire G, and RFV data of this tire T is acquired. Using the acquired RFV data, fluctuation data Dv in the region corresponding to the splice portion Es1 of the cap tread E1 is calculated.
[0037] The image data Ds and variation data Dv acquired in this way are combined into a set and stored in a computer used to build the prediction model PM. Similarly, image data Ds and variation data Dv of the vulcanized tire T of each green tire are acquired and stored for a large number of green tires G. These numerous sets of stored image data Ds and variation data Dv are used as training data.
[0038] If only the cap tread E1 is adopted as the predetermined tire component E, then only the data for the cap tread E1 needs to be used. Therefore, if, for example, the cap tread E1 and side tread E2 are adopted as the predetermined tire component E, then the data for the adopted predetermined tire components E1 and E2 should be used.
[0039] In machine learning using training data, the relationship (correlation) between the features of each image data Ds and the magnitude of the variation data Dv is analyzed. More specifically, using a large number of sets of image data Ds and variation data Dv, the relationship between the pixel value (luminance value) of each pixel in the image data Ds, the distribution of those pixel values (luminance values), and the magnitude of the variation data Dv is understood, and a predictive model PM is constructed. The smaller the magnitude of the variation data Dv, the better the uniformity of the tire T is considered to be. Therefore, the predictive model PM determines that the state of the splice portion Es in the image data Ds acquired when forming the green tire G of tire T, which has a smaller magnitude of variation data Dv, is more appropriate.
[0040] One type of prediction model PM is a classification model that classifies the input image data Ds into multiple classes. When a classification model is used as the prediction model PM, a threshold is set for the size of the shaded area in Figure 7, and this area is divided into multiple classes based on the threshold. Therefore, the variable data Dv (the size of the shaded area in Figure 7) will be classified into one of the multiple classes based on its size. The classification model PM then predicts the size of the variable data Dv based on the input image data Ds, identifies the class to which the prediction result corresponds, and classifies the input image data Ds into that identified class. The suitability of the splice Es is determined based on the class to which the input image data Ds has been classified.
[0041] Specifically, setting Y as the threshold, cases where the area of the shaded portion in Figure 7 is less than or equal to Y are classified as Class A, and cases where it is greater than Y are classified as Class B. In Class A, the area of the shaded portion in Figure 7 is relatively small, so the state of the splice portion Es is judged to be appropriate. In Class B, the area of the shaded portion in Figure 7 is considerably large, so the state of the splice portion Es is judged to be inappropriate.
[0042] When image data Ds of a specified tire component E of the green tire Gt to be inspected is input to the classification model PM, the magnitude of the fluctuation data Dv is predicted, and it is determined whether the predicted magnitude of the fluctuation data Dv (the magnitude of the shaded area in Figure 7) corresponds to class A or class B, and the corresponding class is identified. If the predicted magnitude of the fluctuation data Dv is identified as class A, the image data Ds input to the classification model PM is classified as class A, and the state of the splice Es is determined to be appropriate. If the predicted magnitude of the fluctuation data Dv is identified as class B, the image data Ds input to the classification model PM is classified as class B, and the state of the splice Es is determined to be inappropriate.
[0043] In the explanation above, the classification model PM classifies the magnitude of the variable data Dv into two classes, but there can be multiple classes, for example, three or four. In any case, it is necessary to predetermine which classes the splice Es state is judged to be appropriate and which classes it is judged to be inappropriate. If there are two classes to classify, the judgment of whether the splice Es is appropriate or not becomes very simple.
[0044] The following describes an example of a procedure for determining whether the splice portion Es of a predetermined tire component E is appropriate during the molding process of the green tire Gt to be inspected, using this inspection system 1.
[0045] First, as illustrated in Figures 1 and 2, image data Ds of the tire component E of the green tire Gt to be inspected, which is wrapped around the drum body 6, are acquired by the image acquisition device 2. The acquired image data Ds are input to the calculation device 3. In the calculation device 3, the image data Ds are input as input data to the prediction model PM, and the magnitude of the variation data Dv in the tire T manufactured by vulcanizing the green tire Gt is predicted by the prediction model PM.
[0046] Then, based on the magnitude of the predicted fluctuation data Dv, the prediction model PM determines whether the splice portion Es of the given tire member E is appropriate or not. That is, the judgment result, whether the splice portion Es is appropriate or inappropriate, is output as output data to the output unit 5. If a monitor is used as the output unit 5, the appropriateness of the splice portion Es is displayed in text. If an indicator light is used as the output unit 5, a color indicating the appropriateness of the splice portion Es is displayed. For example, if the splice portion Es is appropriate, blue is lit, and if the splice portion Es is inappropriate, red is lit.
[0047] After image data Ds is acquired by the image acquisition device 2, the judgment result is immediately output to the output unit 5. Therefore, the suitability of the splice portion Es is determined at the molding stage of the green tire Gt to be inspected. The operator can understand the judgment result from the predictive model PM by visually checking the output unit 5. If the splice portion Es is determined to be unsuitable, the operator checks the condition of the splice portion Es and takes appropriate action.
[0048] If the classification model described above is adopted as the predictive model PM, the judgment of whether the splice part Es is appropriate or not becomes very simple if there are two classes to classify. However, if there are three or more classes to classify, the degree of appropriateness of the state of the splice part Es can be understood in more detail. For example, the state of the splice part Es can be understood in detail as "optimal," "appropriate," "inappropriate," or "very inappropriate." Then, depending on the degree of inappropriateness of the state of the splice part Es, it becomes easier for the worker to take more appropriate action on that splice part Es.
[0049] In one embodiment of the tire manufacturing method, a green tire Gt whose splice portion Es is deemed appropriate based on the inspection method for the splice portion Es of the tire member E described above is vulcanized using a known tire vulcanizing apparatus to manufacture a tire T.
[0050] As described above, according to the embodiment of the inspection system 1 and the embodiment of the inspection method using the inspection system 1, at the stage of the molding process of the green tire Gt to be inspected, the predictive model PM determines whether the state of the splice portion Es is appropriate or inappropriate based on the magnitude of the fluctuation of the RFV data described above. Therefore, it is possible to accurately and efficiently detect whether a predetermined tire member E of the green tire Gt to be inspected is spliced without impairing the good uniformity of the tire T manufactured by vulcanizing the green tire Gt.
[0051] According to this embodiment of the tire manufacturing method, a tire T is manufactured by vulcanizing a green tire Gt in which the splice portion Es of a predetermined tire member E is deemed appropriate, thereby obtaining a tire T with excellent uniformity. Furthermore, since the suitability of the splice portion Es is determined at the molding stage of the green tire Gt, appropriate measures can be taken early on if the splice portion Es is deemed inappropriate for the green tire Gt. In other words, it is possible to avoid the waste of continuing production using green tires Gt with an inappropriate splice portion Es, which would result in the production of tires T that do not meet the predetermined uniformity performance. Therefore, using this embodiment of the tire manufacturing method is advantageous in improving the yield (productivity) of tires T.
[0052] When constructing a predictive model PM and acquiring image data Ds during the molding process of the green tire Gt to be inspected, it is possible to use not only the front view image data of the splice portion Es of a given tire member E, but also processed image data Ds. For example, a difference image data Dv can be used, obtained by subtracting the component of the front view image data Dx of the non-splice portion Ex of a given tire member E from the front view image data Ds of the splice portion Es of that tire member E.
[0053] Specifically, when constructing the prediction model PM, the differential image data Dv of a predetermined tire component E of each green tire G is used, and when determining the suitability of the splice portion Es of a predetermined tire component E of the green tire Gt to be inspected, the differential image data Dv of the predetermined tire component E of that green tire Gt is used. Therefore, when constructing the prediction model PM, front view image data Ds and Dx of the splice portion Es and non-splice portion Ex of the predetermined tire component E of each green tire G are acquired by the image acquisition device 2. Then, the differential data Dv created using the acquired image data Ds and Dx is used. When determining the suitability of the splice portion Es, front view image data Ds and Dx of the splice portion Es and non-splice portion Ex of the predetermined tire component E of the green tire Gt to be inspected are acquired by the image acquisition device 2, and the differential data Dv created using that image data Ds and Dx is input to the prediction model PM.
[0054] The differential data Dv may be created by the computing unit 3, or by a computer other than the computing unit 3. In the differential image data Dv, the contrast between the splice portion Es and the non-splice portion Ex is more pronounced, making it easier to clearly understand the state of the splice portion Es. As a result, this is advantageous for improving the accuracy of the prediction model PM's judgment regarding the suitability of the splice portion Es.
[0055] The predetermined tire component E may be at least one of the cap tread E1, side tread E2, or inner liner E3, but it is desirable to include the cap tread E1, which has the greatest influence on the variation of RFV data. Therefore, the predetermined tire component E may be just one type of cap tread E1, or it may be two types of cap tread E1 and side tread E2, two types of cap tread E1 and inner liner E3, or three types of cap tread E1, side tread E2, and inner liner E3.
[0056] It is preferable to construct a predictive model PM for each specification of the green tire G, and then inspect green tires Gt with the same specifications to determine the suitability of the splice portion Es. This is advantageous for improving the accuracy of the predictive model PM's judgment regarding the suitability of the splice portion Es.
[0057] Furthermore, the prediction model PM can be constructed by mixing data from green tires G with specifications that are somewhat similar, even if they are not identical. Then, the green tires Gt with specifications corresponding to the green tires G whose data was used to construct the prediction model PM are subjected to inspection. Somewhat similar specifications refer to cases where the tire size (width and rim diameter, or either width or rim diameter) differs by only one or two steps. In this way, if the green tires G have somewhat similar specifications, the trends in the variation data Dv of the vulcanized tire T will not differ significantly, so the accuracy of the prediction model PM's judgment on the suitability of the splice Es can be maintained at a certain level.
[0058] The green tires Gt subject to inspection are not limited to green tires used to manufacture pneumatic tires T, but may also be green tires used to manufacture types other than pneumatic tires T. [Examples]
[0059] A classification model (predictive model) was constructed to classify 300 passenger car pneumatic tires T of the same specifications into two classes using the difference image Dv described above. This classification model was used to determine the suitability of the splice portion of the cap tread during the molding process of 100 green tires under inspection. The area of the shaded portion exemplified in Figure 7 was used as the variation data Dv, and the value of α was set to 15°. The threshold for classifying this area into two classes was set to a predetermined value Y. The uniformity measurement of the tires manufactured by vulcanizing the 100 green tires under inspection was then performed, and the area of the shaded portion exemplified in Figure 7 was calculated. If the calculated area was less than or equal to the threshold Y, the splice portion was judged to be appropriate; if it was greater than the threshold Y, the splice portion was judged to be inappropriate. The agreement rate between this judgment result and the suitability judgment result of the splice portion by the classification model, i.e., the accuracy rate of the classification model, was checked. As a result, the accuracy rate was approximately 94%, confirming that the judgment accuracy of the classification model was very high.
[0060] This disclosure encompasses the following inventions. Invention 1: By using machine learning with a dataset of image data showing the state of a splice portion of a predetermined tire component obtained during the molding process of a green tire, and variation data showing the degree of variation in RFV data in the region corresponding to the splice portion obtained by uniformity measurement of a tire manufactured by vulcanizing the green tire, a predictive model is constructed that determines the suitability of the splice portion based on the magnitude of the variation in the RFV data, and this model is stored in the computing device. A method for inspecting the splice portion of a tire member, wherein the appropriateness of the splice portion of the green tire to be inspected is determined at the stage of the molding process of the green tire to be inspected, by inputting the image data showing the state of the splice portion of the predetermined tire member of the green tire to be inspected into the prediction model. Invention 2: A method for inspecting a splice portion of a tire member according to Invention 1, wherein when constructing the prediction model, a maximum value phase indicating the maximum value of the RFV data in the region corresponding to the splice portion is identified, and the size of the area of the range protruding to the positive side of the RFV data, enclosed by the RFV data and the line segment connecting the intersection point of the nearby front-to-back phases centered on the maximum value phase and the RFV data, is used as the variation data. Invention 3: A method for inspecting a splice portion of a tire member according to Invention 2, wherein a threshold is set for the size of the area, and the size of the area is divided into a plurality of classes based on this threshold, and the prediction model classifies the size of the area predicted based on the input image data of the green tire to be inspected into one of the plurality of classes, and determines whether the splice portion is suitable or not based on this classified class. Invention 4: A method for inspecting the splice portion of a tire member according to any one of Inventions 1 to 3, wherein when constructing the prediction model and determining the suitability of the splice portion of the green tire to be inspected, the image data used is a difference image data obtained by subtracting the component of the front view image data of the non-splice portion of the predetermined tire member from the front view image data of the splice portion of the predetermined tire member. Invention 5: A method for inspecting the splice portion of a tire member according to any one of inventions 1 to 4, wherein the predetermined tire member is at least one of a cap tread, a side tread, or an inner liner. Invention 6: A method for inspecting the splice portion of a tire member according to any one of inventions 1 to 5, wherein the aforementioned prediction model is constructed for each specification of the green tire. Invention 7: An inspection system for the splice portion of a tire member, comprising an image acquisition device that acquires image data showing the state of the splice portion of a predetermined tire member during the molding process of a green tire, and a calculation device, The computing device stores a predictive model that determines the suitability of the splice portion based on the magnitude of the fluctuation of the RFV data, which is constructed by machine learning using a dataset of the image data and fluctuation data showing the degree of fluctuation of RFV data in the region corresponding to the splice portion, obtained by uniformity measurement of the tire manufactured by vulcanizing the green tire, as training data. A tire component splice inspection system configured such that, by inputting image data showing the state of the splice portion of a predetermined tire component of the green tire to be inspected into the prediction model, the suitability of the splice portion of the green tire to be inspected is determined at the stage of the molding process of the green tire to be inspected. Invention 8: A method for manufacturing a tire, comprising vulcanizing a green tire to be inspected, which is determined to have an appropriate splice portion of a predetermined tire member according to the inspection method for the splice portion of a tire member described in any of Inventions 1 to 6, thereby manufacturing a tire. [Explanation of Symbols]
[0061] 1. Inspection System 2 Image acquisition device 3 Computing device 3a Storage section 3b Arithmetic section 4 Input section 5 Output section 6 drum bodies 6a Center axis G Green Tire Green tires subject to GT inspection E(E1, E2, E3) Tire components Es Splice Ex Non-spliced part T-Tires (Vulcanized Tires) M Standard Mark Image data of the Ds splice section Image data of the non-spliced portion of Dx
Claims
1. By using machine learning with a dataset of image data showing the state of a splice portion of a predetermined tire component obtained during the molding process of a green tire, and variation data showing the degree of variation in RFV data in the region corresponding to the splice portion obtained by uniformity measurement of a tire manufactured by vulcanizing the green tire, a predictive model is constructed that determines the suitability of the splice portion based on the magnitude of the variation in the RFV data, and this model is stored in the computing device. A method for inspecting the splice portion of a tire member, wherein the appropriateness of the splice portion of the green tire to be inspected is determined at the stage of the molding process of the green tire to be inspected, by inputting the image data showing the state of the splice portion of the predetermined tire member of the green tire to be inspected into the prediction model.
2. A method for inspecting a splice portion of a tire member according to claim 1, wherein when constructing the prediction model, a maximum value phase indicating the maximum value of the RFV data in the region corresponding to the splice portion is identified, and the size of the area of the range protruding to the positive side of the RFV data, enclosed by the line segment connecting the intersection point of the nearby front-to-back phases centered on the maximum value phase and the RFV data, and the RFV data, is used as the fluctuation data.
3. A method for inspecting a splice portion of a tire member according to claim 2, wherein a threshold is set for the size of the area, and the size of the area is divided into a plurality of classes based on this threshold, and the prediction model classifies the size of the area predicted based on the input image data of the green tire to be inspected into one of the plurality of classes, and determines whether the splice portion is suitable or not based on this classified class.
4. A method for inspecting the splice portion of a tire member according to claim 1 or 2, wherein when constructing the prediction model and determining whether the splice portion of the green tire to be inspected is suitable, the image data used is a difference image obtained by subtracting the component of the front view image data of the non-splice portion of the predetermined tire member from the front view image data of the splice portion of the predetermined tire member.
5. The method for inspecting the splice portion of a tire member according to claim 1 or 2, wherein the predetermined tire member is at least one of a cap tread, a side tread, or an inner liner.
6. The method for inspecting the splice portion of a tire member according to claim 1 or 2, wherein the prediction model is constructed for each specification of the green tire.
7. An inspection system for the splice portion of a tire member, comprising an image acquisition device that acquires image data showing the state of the splice portion of a predetermined tire member during the molding process of a green tire, and a calculation device, The computing device stores a predictive model that determines the suitability of the splice portion based on the magnitude of the fluctuation of the RFV data, which is constructed by machine learning using a dataset of the image data and fluctuation data showing the degree of fluctuation of RFV data in the region corresponding to the splice portion, obtained by uniformity measurement of the tire manufactured by vulcanizing the green tire, as training data. A tire component splice inspection system configured such that, by inputting image data showing the state of the splice portion of a predetermined tire component of the green tire to be inspected into the prediction model, the suitability of the splice portion of the green tire to be inspected is determined at the stage of the molding process of the green tire to be inspected.
8. A method for manufacturing a tire, comprising vulcanizing a green tire to be inspected, which is determined to have an appropriate splice portion of a predetermined tire member according to the inspection method for the splice portion of a tire member described in any one of claims 1 to 3.