Mounting condition detection device

Abstract

To provide a mounting condition detection device that can accurately determine whether the bobbin is properly mounted or not.SOLUTION: The mounting condition detection device includes: a control unit; and a photographing device configured to take pictures of a bobbin mounted on a bobbin holder of a yarn winder. The control unit executes a process of acquiring an estimation model that is generated by machine-learning a plurality of learning data. Each of the plurality of learning data associates a bobbin image for learning that depicts the bobbin with mounting condition information representing a mounting condition of the bobbin as a label. The controller further performs: a process of acquiring a bobbin image for determination from the photographing device by photographing a bobbin mounted on the bobbin holder; and a process of determining whether the bobbin is normally mounted on the bobbin holder based on the mounting condition information output from the estimation model by inputting the bobbin image for determination into the estimation model.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention primarily relates to a device for detecting the mounting state of a bobbin in a bobbin holder. [Background technology]

[0002] Patent Document 1 discloses a winding machine, a cart, and a paper tube supplying device. Two winding machines are provided side by side. The winding machine winds the yarn to produce a package. The cart pushes and sets paper tubes to each of the two winding machines. The paper tube supplying device supplies the paper tube to the cart. The paper tube supplying device determines the type and orientation of the paper tube by detecting the position of the slit in the paper tube using a photoelectric sensor. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Utility Model Application Publication No. 4-97769 Summary of the Invention [Problem to be solved by the invention]

[0004] The method of detecting slits in a paper tube using a photoelectric sensor as in Patent Document 1 may result in erroneous detection due to, for example, unevenness in the paper tube, gaps between paper tubes, or foreign objects. In addition, erroneous detection may occur not only when detecting slits, but also when detecting various phenomena related to the installation state of the bobbin, such as the size of the gap between bobbins. Therefore, there has been a demand for a configuration that can accurately determine whether the installation state of the bobbin is appropriate.

[0005] The present disclosure has been made in consideration of the above circumstances, and its main objective is to provide an installation state detection device, an installation state detection method, and an installation state detection program that can accurately determine whether the bobbin installation state is appropriate or not. [Means for solving the problem]

[0006] In one example of the present disclosure, an attachment state detection device is provided. The attachment state detection device includes a control device and a photographing device configured to be able to photograph a bobbin attached to a bobbin holder of a yarn winding machine. The control device executes a process of acquiring an estimation model generated by machine learning a plurality of learning data. Each of the plurality of learning data associates, as a label, attachment state information representing an attachment state of the bobbin with a learning bobbin image depicting the bobbin. The control device further executes a process of acquiring a bobbin image for determination from the photographing device by photographing the bobbin attached to the bobbin holder, and a process of inputting the bobbin image for determination into the estimation model and determining whether or not the bobbin is normally attached to the bobbin holder based on the attachment state information output from the estimation model.

[0007] In this mounting state detection device, an estimation model is used in which the relationship between the learning bobbin image and the mounting state information is learned. As a result, the mounting state detection device can obtain the mounting state information of the bobbin by inputting the judgment bobbin image to the estimation model, and can determine the mounting state of the bobbin.

[0008] In one example of the present disclosure, the process of determining whether or not a mounting orientation of the bobbin with respect to the bobbin holder is normal is determined.

[0009] This allows the mounting state detection device to detect specific bobbin mounting abnormalities.

[0010] In one example of the present disclosure, the bobbin has a slit formed therein. The mounting state information as the label includes a bobbin range indicating the range of the bobbin in the learning bobbin image and a slit range indicating the range of the slit in the learning bobbin image. The estimation model receives the judgment bobbin image and outputs the bobbin range in the judgment bobbin image and the slit range in the judgment bobbin image as mounting state information. Whether the mounting orientation is normal or not is determined based on the positional relationship between the bobbin range and the slit range output from the estimation model.

[0011] The mounting state detection device focuses on the positional relationship between the bobbin area and the slit area in the bobbin image, thereby being able to more accurately determine whether the mounting orientation of the bobbin relative to the bobbin holder is normal.

[0012] In one example of the present disclosure, the bobbin holder has a plurality of bobbins arranged side by side. In the determining process, it is determined whether or not the intervals between the bobbins included in the determination bobbin image are normal.

[0013] This allows the mounting state detection device to detect specific bobbin mounting abnormalities.

[0014] In one example of the present disclosure, the mounting state information as the label includes a bobbin range indicating a range of the bobbin in the learning bobbin image. The estimation model receives the judgment bobbin image and outputs the bobbin range for each bobbin included in the judgment bobbin image. Whether or not the interval is normal is determined based on a positional relationship between adjacent bobbin ranges among the bobbin ranges output from the estimation model.

[0015] The attachment state detection device focuses on the spacing between bobbins in the bobbin image, thereby being able to more accurately determine whether the spacing between adjacent bobbins is normal.

[0016] In one example of the present disclosure, the control device further executes a process of notifying an abnormality in the attachment of the bobbin when it is determined that the bobbin is not normally attached to the bobbin holder.

[0017] This allows the mounting state detection device to notify the operator of an abnormality in the mounting state of the bobbin.

[0018] In one example of the present disclosure, the photographing device photographs the bobbin after starting an operation of mounting the bobbin on the bobbin holder and before starting winding of the yarn by the yarn winding machine.

[0019] This allows the mounting state detection device to determine the mounting state of the bobbin before a problem occurs due to incorrect mounting of the bobbin.

[0020] In one example of the present disclosure, a plurality of the bobbins are arranged in the bobbin holder. A plurality of the image capturing devices are arranged along an axial direction of the bobbin holder. The image capturing devices generate the determination bobbin image including one or a plurality of the bobbins.

[0021] As a result, even when a plurality of bobbins are arranged side by side in the bobbin holder, the mounting state of these bobbins can be detected with high accuracy by using a plurality of image capturing devices.

[0022] In one example of the present disclosure, a plurality of the bobbins are arranged in the bobbin holder, and one of the image capturing devices generates the judgment bobbin image including all of the bobbins.

[0023] As a result, even when multiple bobbins are arranged side by side in a bobbin holder, by using a wide-angle camera that can photograph multiple bobbins at once, the number of camera devices can be reduced while the installation status of these bobbins can be accurately detected.

[0024] In one example of the present disclosure, the bobbins are arranged in a plurality of rows in the bobbin holder. The imaging device is configured to be movable along an axial direction of the bobbin holder. After generating the determination bobbin image including the bobbin, the imaging device moves to generate the determination bobbin image including another bobbin.

[0025] As a result, even when multiple bobbins are arranged side by side in the bobbin holder, the installation status of these bobbins can be accurately detected while reducing the number of photographing devices by running the photographing device along the axial direction of the bobbin holder.

[0026] In one example of the present disclosure, the bobbins are inserted into the bobbin holder from an insertion end and slid to arrange the bobbins side by side. The image capturing device captures images of the bobbins inserted into the bobbin holder from the insertion end in sequence to generate the determination bobbin images for all of the bobbins.

[0027] As a result, even when multiple bobbins are arranged side by side in the bobbin holder, the installation status of these bobbins can be accurately detected while reducing the number of photographing devices by sequentially photographing the bobbins as they are inserted.

[0028] In one example of the present disclosure, a plurality of the photographing devices are provided. The photographing devices photograph the bobbin from a plurality of directions. Whether or not the bobbin is properly attached to the bobbin holder is determined based on the plurality of determination bobbin images photographed from the plurality of directions.

[0029] This allows the determination to be made based on the results of photographing a wide range in the circumferential direction of the bobbin, thereby improving the detection accuracy.

[0030] In one example of the present disclosure, the imaging device images the bobbin multiple times while the bobbin is rotating around an axial direction of the bobbin holder to generate multiple bobbin images for determination. Whether or not the bobbin is properly attached to the bobbin holder is determined based on the multiple bobbin images for determination.

[0031] This allows the determination to be made based on the results of photographing the entire circumferential direction of the bobbin, thereby improving the detection accuracy.

[0032] In another example of the present disclosure, an attachment state detection method is provided which is executed by an attachment state detection device. The attachment state detection device includes a photographing device configured to be able to photograph a bobbin attached to a bobbin holder of a yarn winding machine. The attachment state detection method includes a step of acquiring an estimation model generated by machine learning a plurality of learning data. Each of the plurality of learning data associates, as a label, attachment state information representing an attachment state of the bobbin with a learning bobbin image depicting the bobbin. The attachment state detection method further includes a step of acquiring a judgment bobbin image from the photographing device by photographing the bobbin attached to the bobbin holder, and a step of inputting the judgment bobbin image into the estimation model and determining whether or not the bobbin is normally attached to the bobbin holder based on the attachment state information output from the estimation model.

[0033] In this mounting state detection method, an estimation model is used in which the relationship between the learning bobbin image and the mounting state information is learned. In this manner, the mounting state detection method can obtain the mounting state information of the bobbin by inputting the judgment bobbin image to the estimation model, and can judge the mounting state of the bobbin.

[0034] In another example of the present disclosure, an attachment state detection program executed by an attachment state detection device is provided. The attachment state detection device includes a photographing device configured to photograph a bobbin attached to a bobbin holder of a yarn winding machine. The attachment state detection program causes the attachment state detection device to execute a step of acquiring an estimation model generated by machine learning a plurality of learning data. Each of the plurality of learning data associates, as a label, attachment state information representing an attachment state of the bobbin with a learning bobbin image depicting the bobbin. The attachment state detection program further causes the attachment state detection device to execute a step of acquiring a bobbin image for determination from the photographing device by photographing the bobbin attached to the bobbin holder, and a step of inputting the bobbin image for determination into the estimation model and determining whether or not the bobbin is normally attached to the bobbin holder based on the attachment state information output from the estimation model.

[0035] In this installation state detection program, an estimation model is used in which the relationship between the learning bobbin image and the installation state information is learned. As a result, the installation state detection program can obtain the bobbin installation state information by inputting the judgment bobbin image to the estimation model, and can judge the installation state of the bobbin. [Brief description of the drawings]

[0036] [Figure 1] FIG. 2 is a front view of the yarn winding machine according to the embodiment. [Diagram 2] FIG. 2 is a side view of the yarn winding machine. [Diagram 3] FIG. 2 is a block diagram of the yarn winding machine. [Figure 4] FIG. 1 is a diagram illustrating a method for constructing an estimation model by machine learning. [Diagram 5] FIG. 1 illustrates a method for obtaining estimated data using an estimation model. [Figure 6] FIG. 13 is a diagram showing an estimation result of the estimation model superimposed on a bobbin image. [Figure 7]FIG. 11 is a flowchart showing a process for determining whether the bobbin attachment state is appropriate. [Figure 8] FIG. 2 is a diagram illustrating an example of a hardware configuration of a control device. [Figure 9] FIG. 1 is a diagram illustrating an example of a learning dataset. [Figure 10] FIG. 2 is a diagram illustrating an example of a functional configuration of a control device. [Figure 11] FIG. 2 is a diagram conceptually illustrating a learning process performed by a learning unit. [Figure 12] 11 is a diagram conceptually showing a mounting state determination process performed by a determination unit. FIG. [Figure 13] FIG. 13 is a diagram showing a device configuration of a mounting state determination system in a second modified example. [Figure 14] FIG. 13 is a diagram showing an example of a learning dataset in a third modified example. [Figure 15] FIG. 13 is a diagram conceptually showing a learning process by a learning unit in a third modified example. [Figure 16] 13A and 13B are diagrams conceptually showing a process of determining whether or not the device is being worn by a determining unit in a third modified example. [Figure 17] FIG. 13 is a side view of a yarn winding machine according to a fourth modified example. [Figure 18] FIG. 13 is a side view of a yarn winding machine according to a fifth modified example. [Figure 19] FIG. 13 is a side view of a yarn winding machine in a sixth modified example. [Figure 20] FIG. 13 is a front view of a yarn winding machine according to a seventh modified example. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0037] Hereinafter, each embodiment according to the present invention will be described with reference to the drawings. In the following description, the same parts and components are given the same reference numerals. Their names and functions are also the same. Therefore, detailed description thereof will not be repeated. Note that each embodiment and each modified example described below may be appropriately and selectively combined.

[0038] <A. Overview> First, an overview of the wearing state detection device according to the embodiment will be described.

[0039] The wearing state detection device determines whether the bobbin is properly mounted based on the bobbin image obtained from the imaging device by imaging the bobbin mounted on the bobbin holder of the winding machine. In order to realize such a determination function, first, the wearing state detection device acquires an estimation model generated by machine learning a plurality of learning data.

[0040] Each of the plurality of learning data is associated with a learning bobbin image showing the bobbin, with the wearing state information representing the wearing state of the bobbin as a label. The "wearing state information" referred to here means information that can directly or indirectly identify the wearing state of the bobbin with respect to the bobbin holder. In other words, the wearing state information is information that correlates with the wearing state of the bobbin with respect to the bobbin holder. Specific examples of the wearing state information will be described later.

[0041] The wearing state detection device causes the imaging device to image the bobbin mounted on the bobbin holder to obtain a determination bobbin image from the imaging device, and inputs the determination bobbin image into the above-mentioned estimation model. Next, the wearing state detection device determines whether the bobbin is properly mounted on the bobbin holder based on the wearing state information output from the estimation model. Thereby, the wearing state detection device can accurately determine the wearing state of the bobbin with respect to the bobbin holder.

[0042] <B. Winding Machine 1> Next, the embodiments of the present disclosure will be described in more detail with reference to the drawings. FIG. 1 is a front view of a winding machine 1 according to an embodiment of the present disclosure. FIG. 2 is a side view of the winding machine 1. FIG. 3 is a block diagram of the winding machine 1. In the following description, the upstream or downstream in the running direction of the yarn may be simply referred to as upstream or downstream.

[0043] 1, a spinning machine (not shown) is disposed upstream of the yarn winding machine 1. The spinning machine produces a yarn 93 and supplies it to the yarn winding machine 1. The yarn winding machine 1 winds the yarn 93 onto a bobbin 91 to produce a package 94.

[0044] The bobbin 91 is a cylindrical member. As shown in FIG. 2, a slit 91a is formed in the bobbin 91. The slit 91a is a groove formed in the surface of the bobbin 91 along the circumferential direction. The slit 91a may be formed over the entire circumference, or may be formed only on a part of the circumference. The slit 91a is formed at a position toward one side in the axial direction of the bobbin 91. In other words, since the slit 91a has an asymmetric shape with respect to the center in the axial direction, it is necessary to position the bobbin 91 in the correct mounting orientation.

[0045] As shown in Fig. 1, a yarn 93 is wound around a bobbin 91 at the upper winding position to form a yarn layer and a package 94. On the other hand, no yarn 93 is wound around a bobbin 91 at the lower standby position. The yarn 93 may be a synthetic yarn such as nylon or polyester. The yarn 93 may also be an elastic yarn such as spandex.

[0046] As shown in FIG. 2, in this embodiment, a plurality of yarns 93 aligned in the axial direction of the first bobbin holder 41 (second bobbin holder 42) are supplied from the spinning machine to the yarn winding machine 1 through the yarn feed roller 100. A plurality of bobbins 91 are arranged in the axial direction of the first bobbin holder 41 (second bobbin holder 42). The yarn winding machine 1 winds the plurality of yarns 93 onto the bobbins 91, respectively, to manufacture a plurality of packages 94. Hereinafter, in the axial direction of the first bobbin holder 41 (second bobbin holder 42), the side of a turret plate 40 (described later) is referred to as a second side, and the opposite side is referred to as a first side. When the bobbin 91 is arranged in the correct mounting orientation, the slit 91a is located on the first side.

[0047] The following describes in detail the yarn winding machine 1. As shown in Fig. 1, the yarn winding machine 1 includes a frame 11, a first housing 20, a second housing 30, and a turret plate (bobbin holder moving mechanism) 40.

[0048] The frame 11 is a member that holds each component included in the yarn winding machine 1. A first housing 20 and a second housing 30 are attached to the frame 11. The first housing 20 and the second housing 30 are movable up and down relative to the frame 11.

[0049] A traverse device 21 is attached to the first housing 20. The traverse device 21 traverses the yarn 93 sent downstream by reciprocating along the axial direction of a first bobbin holder 41 (described later) to the winding width of a package 94 with a traverse guide 23 (described later) engaged with the yarn 93. This traverse motion of the yarn 93 forms a yarn layer on the bobbin 91 or the package 94. As shown in FIG. 3, the traverse device 21 includes a traverse cam 22 and a traverse guide 23.

[0050] The traverse cam 22 is a roller-shaped member disposed in parallel to the bobbin 91 or the package 94. A spiral cam groove is formed on the outer circumferential surface of the traverse cam 22. The traverse cam 22 is rotated by a traverse motor 51.

[0051] The traverse motor 51 is controlled by a control device 50, which will be described later. The traverse guide 23 is a part that engages with the yarn 93. The tip of the traverse guide 23 has, for example, a substantially U-shaped guide portion, and engages with the yarn 93 by pinching the yarn 93 in the winding width direction. The base end of the traverse guide 23 is positioned in the cam groove of the traverse cam 22. With this configuration, the traverse guide 23 can be reciprocated in the winding width direction by driving the traverse cam 22 to rotate.

[0052] A contact roller 31 is rotatably attached to the second housing 30. The contact roller 31 rotates while contacting the yarn layer of the package 94 with a predetermined pressure during winding of the yarn 93, thereby feeding the yarn 93 from the traverse guide 23 to the yarn layer of the package 94 and adjusting the shape of the yarn layer of the package 94. The contact roller 31 may be rotationally driven by a driving unit such as a motor.

[0053] The second housing 30 is provided with an operation panel 32. The operation panel 32 is a device that is operated by an operator. The operator issues instructions to the yarn winding machine 1 by operating the operation panel 32. Examples of instructions issued by the operator include starting winding, stopping winding, changing winding conditions, etc.

[0054] The turret plate 40 is a disk-shaped member. The turret plate 40 is rotatably attached to the frame 11. The turret plate 40 is rotatable about a normal line passing through the center of the disk as a rotation axis. The turret plate 40 is rotationally driven by a turret motor 53 shown in FIG. 3. The turret motor 53 is controlled by a control device 50 described later.

[0055] The turret plate 40 is provided with a first bobbin holder 41 and a second bobbin holder 42 at two locations facing each other across the center of the disk. A plurality of bobbins 91 can be mounted on the first bobbin holder 41, lined up in the axial direction of the first bobbin holder 41. A plurality of bobbins 91 can be mounted on the second bobbin holder 42, lined up in the axial direction of the second bobbin holder 42. The positions of the first bobbin holder 41 and the second bobbin holder 42 can be changed by rotating the turret plate 40. Note that, as long as the positions of the first bobbin holder 41 and the second bobbin holder 42 can be changed, another device may be used instead of the turret plate 40.

[0056] The first bobbin holder 41 is rotatable relative to the turret plate 40 around the axial position of the first bobbin holder 41 as the center of rotation. The first bobbin holder 41 is rotationally driven by a first bobbin holder motor 54 shown in FIG. 3. Similarly, the second bobbin holder 42 is rotatable relative to the turret plate 40 around the axial position of the second bobbin holder 42 as the center of rotation. The second bobbin holder 42 is rotationally driven by a second bobbin holder motor 55 shown in FIG. 3. The first bobbin holder motor 54 and the second bobbin holder motor 55 are controlled by a control device 50 described later.

[0057] Hereinafter, the first bobbin holder 41 and the second bobbin holder 42 will be collectively referred to as the bobbin holders 41, 42. Fig. 1 shows the bobbin holders 41, 42 lined up one above the other. At this time, the position of the higher bobbin holder 41, 42 is the winding position, and the position of the lower bobbin holder 41, 42 is the standby position. The yarn winding machine 1 winds the yarn 93 onto the bobbin 91 of the bobbin holder 41, 42 that is in the winding position to produce a package 94.

[0058] Furthermore, when a predetermined amount of yarn 93 has been wound and the package 94 of the first bobbin holder 41 is full, the turret plate 40 rotates, switching the positions of the first bobbin holder 41 and the second bobbin holder 42. After that, the package 94 of the first bobbin holder 41, which is now full and in the standby position, is collected, and the yarn 93 is wound around the bobbin 91 of the second bobbin holder 42, which is in the winding position. A new bobbin 91 is attached to the first bobbin holder 41 from which the package 94 has been collected.

[0059] The control device 50 includes a control unit 50a, a learning unit 50b, a determination unit 50c, and a storage unit 50d. Specifically, the control device 50 is configured as a known computer, and includes a central processing unit (CPU), a random access memory (RAM), a solid state drive (SSD), and the like. The CPU is a type of processor. The SSD stores programs and data for controlling the yarn winding machine 1 in advance. The CPU reads the programs into the RAM and executes them, thereby allowing the control device 50 to operate as the control unit 50a, the learning unit 50b, and the determination unit 50c. The SSD corresponds to the storage unit 50d. Note that instead of the SSD, a hard disk drive (HDD) or a flash memory may be used. Alternatively, a storage provided outside the control device 50 and capable of communicating with the control device 50 may be used as the storage unit.

[0060] The control unit 50a is responsible for the overall control performed by the control device 50. The control unit 50a processes data input from the outside to the control device 50 or data stored in the memory unit 50d. The control unit 50a stores the data obtained by this processing in the memory unit 50d or outputs it to the outside of the control device 50. The learning unit 50b performs a process of constructing an estimation model using machine learning. The details of the estimation model constructed by the learning unit 50b will be described later. The judgment unit 50c performs a judgment process based on the data input from the outside to the control device 50 and the estimation model constructed by the learning unit 50b and stored in the memory unit 50d. The memory unit 50d stores data according to the processing of the control unit 50a.

[0061] FIG. 3 shows the photographing device 61 and the notification unit 65. The photographing device 61 photographs the bobbin 91 to generate a bobbin image. The photographing device 61 is, for example, a camera that captures still images. However, the photographing device 61 may be a video camera that captures moving images. If the photographing device 61 is a video camera, it is sufficient for the photographing device 61 to generate an image from the moving image. The notification unit 65 notifies the operator of the determination result of the determination unit 50c. The notification unit 65 is, for example, a notification lamp or a notification buzzer.

[0062] 3 shows the attachment state detection device 10. The attachment state detection device 10 detects the attachment state of the bobbin 91 to the bobbin holder 41 of the yarn winding machine 1. The attachment state detection device 10 includes a control device 50, a photographing device 61, and an informing unit 65. The control device 50, as a part constituting the attachment state detection device 10, performs the following processes.

[0063] That is, the control unit 50a outputs data to the learning unit 50b and the determination unit 50c to perform processing. The control unit 50a stores data obtained by the learning unit 50b and the determination unit 50c performing processing in the storage unit 50d. The control unit 50a also stores data (bobbin image) captured by the imaging device 61 and generated in the storage unit 50d. The control unit 50a controls and operates the notification unit 65 to output the detected mounting state. Note that further examples of the control target of the control unit 50a include the traverse motor 51, the turret motor 53, the first bobbin holder motor 54, and the second bobbin holder motor 55. The learning unit 50b performs machine learning based on the bobbin image and the like stored in the storage unit 50d to construct an estimation model. The determination unit 50c performs a determination process regarding the mounting state of the bobbin 91 based on data input from the outside to the control device 50 and the estimation model constructed by the learning unit 50b and stored in the storage unit 50d. The determination result obtained by the determination process of the determination unit 50c is output to the outside by the control unit 50a or stored in the storage unit 50d. As described above, the storage unit 50d stores data obtained by the processes of the control unit 50a, the learning unit 50b, and the determination unit 50c.

[0064] The mounting orientation of the bobbin 91 will be briefly described. As described above, the bobbin 91 has the slit 91a formed therein, so the first and second axial sides of the bobbin 91 are not symmetrical. Therefore, the bobbin 91 needs to be mounted in the bobbin holders 41, 42 in the correct orientation. Specifically, the bobbin 91 needs to be attached to the bobbin holders 41, 42 in an orientation in which the slit 91a is positioned on the opposite side to the turret plate 40.

[0065] Next, the gap between adjacent bobbins 91 will be briefly described. As described above, in this embodiment, a plurality of bobbins 91 are arranged in the axial direction of the first bobbin holder 41 (second bobbin holder 42), and the thread 93 is wound around each of the bobbins 91. Therefore, the traverse guide 23 and thread guides (not shown) are also arranged in the axial direction. Here, if the gap between the bobbins 91 falls outside of an appropriate range, the position of the bobbin 91 falls outside of the appropriate range. As a result, the difference between the positions of the traverse guide 23, etc. and the bobbin 91 becomes large, and the thread 93 cannot be wound appropriately.

[0066] The mounting state detection device 10 may detect only one of the mounting orientation of the bobbin 91 and the gap between adjacent bobbins 91. Alternatively, the mounting state detection device 10 may detect the mounting state of the bobbin 91 other than these.

[0067] As shown in Figs. 1 and 2, the photographing device 61 is disposed so that the lens faces the bobbin holders 41, 42. In this embodiment, the photographing device 61 photographs the bobbin 91 in the standby position, but instead of or in addition to that, it may photograph the bobbin 91 in the winding position. As shown in Fig. 2, one photographing device 61 photographs two bobbins 91. However, the photographing device 61 may be configured to photograph one or three or more bobbins 91. An image generated by the photographing device 61 photographing the bobbin 91 is referred to as a bobbin image.

[0068] The photographing device 61 photographs the bobbin 91 to generate a plurality of bobbin images while the second bobbin holder 42 is being rotated by the second bobbin holder motor 55. As described above, the bobbin 91 may have the slit 91a formed only in a part of the circumferential direction. Therefore, in a situation where the part where the slit 91a is formed is in the blind spot of the photographing device 61, the slit 91a may not be photographed. In this regard, by photographing the bobbin 91 while rotating the second bobbin holder 42, various parts of the outer circumferential surface of the bobbin 91 can be photographed, so that the bobbin image including the slit 91a can be generated more reliably.

[0069] FIG. 4 conceptually illustrates a process for the learning unit 50b to construct an estimation model using machine learning. The estimation model is constructed by machine learning of a bobbin image as learning data and answer data indicating the range of the bobbin and the range of the slit in the bobbin image. Therefore, this machine learning is supervised learning. Although a specific method for performing machine learning is arbitrary, in this embodiment, deep learning is used. Thereby, the learning unit 50b can learn the association between the bobbin image and the range of the bobbin 91, and extract features for estimating the range of the bobbin 91 from the bobbin image. Similarly, the learning unit 50b can learn the association between the bobbin image and the range of the slit 91a, and extract features for estimating the range of the slit 91a from the bobbin image. In this way, the learning unit 50b constructs an estimation model. The estimation model constructed by the learning unit 50b is stored in the storage unit 50d.

[0070] As shown in Fig. 5, the determination unit 50c generates estimated data indicating the range of the bobbin and the range of the slit in the bobbin image from the bobbin image and the estimated model generated by the photographing device 61. The estimation result of the estimated model is shown in Fig. 6. Fig. 6 is a diagram in which a bobbin range 81 and a slit range 82 are superimposed on the bobbin image.

[0071] The above-described estimation model is an example, and the estimation model may be constructed by another method. That is, instead of deep learning, machine learning may be performed by a user specifying a feature amount related to the contour or the like of the bobbin 91. Also, the estimation model is an example, and for example, the orientation of the bobbin 91 may be used as the estimation model.

[0072] The storage unit 50d stores the correct mounting orientation of the bobbin 91. In this embodiment, the correct mounting orientation of the bobbin 91 is the orientation in which the slit 91a is located on the first side. The storage unit 50d also stores a threshold range that is an appropriate range for the gap between adjacent bobbins 91. The threshold range is determined in advance experimentally or empirically.

[0073] The determination unit 50c determines whether the mounting state of the bobbin 91 is appropriate. In detail, the determination unit 50c creates a central virtual line 81a indicating the central position of the axial direction of the bobbin range 81 based on the bobbin range 81 by the estimation model. The determination unit 50c accesses the storage unit 50d to obtain the correct mounting orientation of the bobbin 91. Next, the determination unit 50c determines whether the slit range 82 is on the first side or the second side of the axial direction from the central virtual line 81a. If the slit range 82 is on the first side from the central virtual line 81a, the determination unit 50c determines that the mounting orientation of the bobbin 91 is correct. Note that, if the determination unit 50c has a storage device, the correct mounting orientation of the bobbin 91 may be stored in the storage device of the determination unit 50c.

[0074] Next, the determination unit 50c compares the two adjacent bobbin ranges 81 and calculates the length L1, which is the gap between the bobbin ranges 81. The determination unit 50c accesses the storage unit 50d to obtain a threshold range, which is an appropriate range for the gap between the adjacent bobbins 91. Next, the determination unit 50c determines whether the length L1 satisfies the threshold range. If the length L1 calculated from the bobbin image satisfies the threshold range, the determination unit 50c determines that the gap between the bobbin ranges 81 is appropriate. Note that FIG. 6 is a diagram for easily illustrating the processing of the determination unit 50c, and the processing of superimposing the bobbin range 81 and the slit range 82 on the bobbin image may be omitted. Note that, if the determination unit 50c has a storage device, the threshold range may be stored in the storage device of the determination unit 50c.

[0075] Next, the process performed by the mounting state detection device 10 will be described mainly with reference to FIG.

[0076] The control device 50 determines whether it is time to detect the bobbin attachment state (S101). The detection timing is after the start of the operation of attaching the bobbin 91 to the bobbin holders 41, 42 and before the start of winding the yarn 93 by the yarn winding machine 1. The control device 50 may determine whether it is time to detect based on a detection value of a sensor, or may determine that it is time to detect when an instruction from an operator is received.

[0077] When the control unit 50a of the control device 50 determines the detection timing, it instructs the imaging device 61 to image the bobbin 91 (S102). Next, as described above, the determination unit 50c of the control device 50 uses the bobbin image generated by the imaging device 61 and the estimation model stored in the storage unit 50d to estimate the range of the bobbin 91 and the range of the slit 91a in the bobbin image (S103).

[0078] Next, as described above, the determination unit 50c of the control device 50 determines whether the mounting state of the bobbin 91 is appropriate based on the estimation result of the determination unit 50c (S104). When the control unit 50a of the control device 50 determines that the mounting state of the bobbin 91 is appropriate, it returns to the process of step S101 again. When the control unit 50a of the control device 50 determines that the mounting state of the bobbin 91 is not appropriate, it controls the notification unit 65 to notify the operator of the abnormality (S105).

[0079] By repeating the above processing, the mounting state detection device 10 can detect whether the mounting state of the bobbin 91 is appropriate. In particular, the detection accuracy can be improved by performing the detection using the estimation model constructed by machine learning.

[0080] <C. Hardware Configuration of Control Device 50> Next, with reference to FIG. 8, the hardware configuration of the control device 50 shown in FIG. 3 will be described. FIG. 8 is a diagram showing an example of the hardware configuration of the control device 50.

[0081] The control device 50 includes the above-described storage unit 50d (see FIG. 3), a processor 101, a communication interface 104, a display interface 105, and an input interface 107. These components are connected to a bus 115. Examples of the storage unit 50d include a ROM (Read Only Memory) 102, a RAM 103, and an auxiliary storage device 120.

[0082] The processor 101 is, for example, configured by at least one integrated circuit. The integrated circuit may be, for example, configured by at least one CPU, at least one GPU (Graphics Processing Unit), at least one ASIC (Application Specific Integrated Circuit), at least one FPGA (Field Programmable Gate Array), or a combination thereof.

[0083] The processor 101 controls the operation of the control device 50 by executing various programs. Based on receiving an execution command for various programs, the processor 101 reads the program to be executed from the auxiliary storage device 120 or the ROM 102 to the RAM 103. The RAM 103 functions as a working memory and temporarily stores various data required for the execution of the program.

[0084] A LAN (Local Area Network), an antenna, etc. are connected to the communication interface 104. The control device 50 exchanges data with an external device via the communication interface 104. The external device includes, for example, a server.

[0085] A display device 106 is connected to the display interface 105. The display interface 105 sends an image signal for displaying an image to the display device 106 in accordance with a command from the processor 101 or the like. The display device 106 is, for example, a liquid crystal display, an organic EL (Electro Luminescence) display, or other displays. The display device 106 may be configured integrally with the control device 50, or may be configured separately from the control device 50.

[0086] An input device 108 is connected to the input interface 107. The input device 108 is, for example, a mouse, a keyboard, a touch panel, or other device capable of accepting a user's operation. The input device 108 may be configured integrally with the control device 50, or may be configured separately from the control device 50.

[0087] The auxiliary storage device 120 is, for example, a storage medium such as a hard disk, a flash memory, or an SSD. The auxiliary storage device 120 stores, for example, a learning dataset 122, the above-mentioned estimation model 124, a learning program 126, and a mounting state detection program 128. The storage location of these is not limited to the auxiliary storage device 120, and may be stored in a storage area (for example, a cache memory, etc.) of the processor 101, the ROM 102, the RAM 103, an external device (for example, a server), etc.

[0088] The learning program 126 is a program for generating the estimation model 124 using the learning dataset 122. The learning program 126 may be provided not as a standalone program but as part of an arbitrary program. In this case, the learning process by the learning program 126 is realized in cooperation with the arbitrary program. Even if the program does not include such a part of the modules, it does not deviate from the purpose of the learning program 126 according to the present embodiment. Furthermore, some or all of the functions provided by the learning program 126 may be realized by dedicated hardware. Furthermore, the control device 50 may be configured in a form like a so-called cloud service in which at least one server executes part of the processing of the learning program 126.

[0089] The mounting state detection program 128 is a program for determining the mounting state of the bobbin 91 on the bobbin holders 41 and 42 of the bobbin winder 1 using the learned estimation model 124. The mounting state detection program 128 may be provided not as a single program but incorporated into a part of any program. In this case, the determination process by the mounting state detection program 128 is realized in cooperation with any program. Even a program that does not include such a part of the module does not deviate from the gist of the mounting state detection program 128 according to the present embodiment. Further, part or all of the functions provided by the mounting state detection program 128 may be realized by dedicated hardware. Further, the control device 50 may be configured in a form such as a so-called cloud service in which at least one server executes a part of the processing of the mounting state detection program 128.

[0090] <D. Learning dataset 122> Next, with reference to FIG. 9, the learning dataset 122 shown in FIG. 8 will be described. FIG. 9 is a diagram showing an example of the learning dataset 122.

[0091] The learning dataset 122 includes a plurality of learning data 123. The number of learning data 123 included in the learning dataset 122 is arbitrary. As an example, the number of learning data 123 is several tens to several hundreds of thousands.

[0092] Each of the learning data 123 associates mounting state information representing the mounting state of the bobbin 91 with the learning bobbin image capturing the bobbin 91 as a label (correct value). FIG. 9 shows a combination of classification information and range information as an example of the mounting state information. One or more combinations of classification information and range information are associated with one learning bobbin image.

[0093] The classification information is information that defines the type of object shown in the bobbin image. The classification information is indicated by, for example, identification information that uniquely indicates the type of object. The identification information is predefined for each object. As an example, the identification information "1" indicates the object type "bobbin", and the identification information "2" indicates the object type "slit".

[0094] The range information is information that defines the range of the object in the bobbin image in coordinate information. The coordinate information is defined by, for example, the center coordinates of the object, the horizontal width of the object, and the vertical width of the object.

[0095] <E. Functional Configuration of Control Device 50> Next, with reference to FIGS. 10 to 12, the functional configuration of the control device 50 will be described. FIG. 10 is a diagram showing an example of the functional configuration of the control device 50.

[0096] As shown in FIG. 10, the control device 50 includes a learning unit 50b and a determination unit 50c as functional configurations. These functional configurations will be described in order below.

[0097] (E1. Learning Unit 50b) First, with reference to FIG. 11, the function of the learning unit 50b shown in FIG. 10 will be described. FIG. 11 is a diagram conceptually showing the learning process by the learning unit 50b.

[0098] The learning unit 50b executes a learning process using the above-described learning dataset 122 (see FIG. 9) and generates an estimation model 124 that outputs the type of object shown in the bobbin image and the range of the object shown in the bobbin image. The machine learning algorithm for realizing the detection of the object type and the object range is not particularly limited. Examples of the machine learning algorithm include R-CNN (Regions with Convolutional Neural Network), Fast R-CNN, Faster R-CNN, SSD (Single Shot MultiBox Detector), YOLO (You Only Look Once), and the like. Hereinafter, an example of the machine learning algorithm will be described.

[0099] The estimation model 124 is configured to receive an input of a training bobbin image defined in the training data 123. The training bobbin image is divided into a plurality of regions according to a predetermined setting. Hereinafter, each divided region in the bobbin image is also referred to as a "grid cell." In the example of FIG. 11, the training bobbin image is divided into 35 (=7×5) grid cells.

[0100] The estimation model 124 is composed of, for example, a classification model 125A and a range prediction model 125B. The classification model 125A and the range prediction model 125B may have different network structures, or may be configured to share a part of the network structure.

[0101] The classification model 125A receives a training bobbin image and outputs the type of object contained in the training bobbin image as an estimation result. The number of classification information items output as estimation results is proportional to the number of objects to be classified and the number of grid cells in the training bobbin image. As an example, if the number of objects to be classified is two, a bobbin and a slit, and the number of grid cells is 35 (=7×5), the classification model 125A outputs 70 (=2×35) scores as classification information.

[0102] The range prediction model 125B receives a learning bobbin image as input and outputs range information of objects contained in the learning bobbin image as an estimation result. The range information is defined by a combination of "cx" and "cy" indicating the center coordinates of the object, "w" indicating the horizontal width of the object, and "h" indicating the vertical width of the object. The number of pieces of range information output as an estimation result is proportional to the number of grid cells in the learning bobbin image. As an example, if the learning bobbin image is divided into 35 (=7×5) grid cells, the range prediction model 125B outputs 35 pieces of range information.

[0103] Depending on the machine learning algorithm employed, one or more bounding boxes may be set for each grid cell. The bounding box means a rectangular frame that indicates the approximate range of each object in the bobbin image. In this case, the number of classification information output by the classification model 125A is not only proportional to the number of objects to be classified and the number of grid cells in the learning bobbin image, but also proportional to the number of bounding boxes that are set. In addition, the number of range information output by the range prediction model 125B is not only proportional to the number of grid cells, but also proportional to the number of bounding boxes that are set.

[0104] Next, the process of updating the internal parameters of the estimation model 124 by the learning unit 50b will be described.

[0105] The learning unit 50b inputs the learning bobbin image defined in the learning data 123 to the estimation model 124. As a result, the learning unit 50b acquires classification information and range information as an estimation result of the estimation model 124. Thereafter, the learning unit 50b refers to the wearing state information as a correct value defined in the learning data 123, and calculates an error between the wearing state information and the estimation result of the estimation model 124. Next, the learning unit 50b updates the internal parameters (e.g., weights and biases) of the estimation model 124 so as to reduce the error. The update of the internal parameters is realized, for example, by an error backpropagation method.

[0106] The learning unit 50b repeatedly performs an update process of the internal parameters of the estimation model 124 for each piece of learning data 123 included in the learning data set 122. The estimation model 124A comes to output accurate estimation results as the learning progresses. Through the above learning process, the estimation model 124 receives an input of a bobbin image, and comes to output a score indicating the possibility that a classification target is captured and range information indicating coordinate information of the classification target for each grid cell.

[0107] The learning unit 50b does not need to use all of the learning data 123 included in the learning dataset 122 for the learning process, and may generate the estimation model 124 using a portion of the learning data 123 included in the learning dataset 122. The remaining learning data 123 is used, for example, for evaluation of the estimation model 124.

[0108] (E2. Judgment section 50c) Next, the function of the determination unit 50c shown in Fig. 10 will be described with reference to Fig. 12. Fig. 12 is a diagram conceptually showing the mounting state determination process performed by the determination unit 50c.

[0109] First, the determination unit 50c acquires the estimation model 124 generated by the learning unit 50b from a storage location. The estimation model 124 may be acquired from the storage unit 50d described above or from an external device.

[0110] Next, the determination unit 50c acquires a bobbin image for determination from the photographing device 61 by photographing the bobbin 91 mounted on the bobbin holders 41 and 42. The bobbin image for determination is an image used to determine whether or not the bobbin 91 is normally mounted on the bobbin holders 41 and 42. The determination unit 50c inputs the acquired bobbin image for determination to the estimation model 124 and acquires mounting state information output from the estimation model 124. Thereafter, the determination unit 50c judges whether or not the bobbin 91 is normally mounted on the bobbin holders 41 and 42 based on the mounting state information. In the example of FIG. 12, the above-mentioned classification information and the above-mentioned range information are shown as mounting state information output from the estimation model 124.

[0111] The determination unit 50c integrates the classification information and range information output from the estimation model 124 to detect the presence or absence of a classification target object in the estimation image and range information of the object. More specifically, when the classification target objects are two types, the bobbin 91 and the slit 91a, the classification information specifies a score indicating the possibility that the bobbin 91 exists and a score indicating the possibility that the slit 91a exists for each grid cell. The determination unit 50c determines that the bobbin 91 is included in a grid cell in which the score related to the bobbin 91 exceeds a predetermined value. The determination unit 50c also determines that the slit 91a is included in a grid cell in which the score related to the slit 91a exceeds a predetermined value. Thereafter, the determination unit 50c refers to the range information corresponding to those grid cells to identify a bobbin range indicating the range of the bobbin 91 in the judgment bobbin image and a slit range indicating the range of the slit 91a in the judgment bobbin image.

[0112] The determination unit 50c detects various mounting abnormalities of the bobbin 91 based on the identified bobbin range and slit range.

[0113] As an example, the determination unit 50c determines whether the mounting orientation of the bobbin 91 relative to the bobbin holders 41, 42 is normal. Whether the mounting orientation of the bobbin 91 is normal is determined, for example, based on the positional relationship between the bobbin range and the slit range. More specifically, the determination unit 50c determines whether the relative position of the slit range based on the bobbin range is close to the above-mentioned first side (see FIG. 6) or close to the above-mentioned second side (see FIG. 6). When the position of the slit 91a relative to the bobbin 91 is on the first side, the determination unit 50c determines that the mounting orientation of the bobbin 91 is normal. On the other hand, when the position of the slit 91a relative to the bobbin 91 is on the second side, the determination unit 50c determines that the mounting orientation of the bobbin 91 is abnormal.

[0114] As another example, the determination unit 50c determines whether the intervals between the respective bobbins 91 included in the bobbin image for determination are normal. Whether the intervals are normal is determined, for example, based on the positional relationship between adjacent bobbin ranges within each bobbin range in the bobbin image for determination. More specifically, the determination unit 50c calculates the distance between adjacent bobbin ranges, and based on the distance, calculates the above-described length L1 (see FIG. 6), which is the interval between adjacent bobbins 91. When the length L1 is included in a predetermined threshold range, the determination unit 50c determines that the interval between adjacent bobbins 91 is normal. On the other hand, when the length L1 is not included in the predetermined threshold range, the determination unit 50c determines that the interval between adjacent bobbins 91 is abnormal.

[0115] Note that the output mode of the determination result by the determination unit 50c is not particularly limited. As an example, the determination unit 50c outputs a warning including the type of mounting abnormality that has occurred. The warning may be displayed as a message on the above-described display device 106, may be output as sound, or may be stored in the storage unit 50d as a log.

[0116] <F. First Modified Example> Next, with reference to FIG. 13, a first modified example of the above-described embodiment will be described. In the following description, members that are the same as or similar to those in the foregoing embodiment may be denoted by the same reference numerals in the drawings, and the description thereof may be omitted.

[0117] In the example of FIG. 11 described above, the learning unit 50b and the determination unit 50c were implemented in the same winding machine 1. However, the learning unit 50b and the determination unit 50c do not necessarily have to be implemented in the same winding machine 1. As an example, the learning unit 50b may be implemented in a different device.

[0118] FIG. 13 is a diagram showing an example of the device configuration of the wearing state determination system 500 in this modified example. As shown in FIG. 13, the wearing state determination system 500 includes one or more winding machines 1 and one or more information processing devices 200. In the example of FIG. 13, the wearing state determination system 500 is composed of three winding machines 1A to 1C and one information processing device 200.

[0119] The winding machines 1A to 1C and the information processing device 200 are connected to the same network NW and are configured to be able to communicate with each other. The winding machine 1 and the information processing device 200 may be communicatively connected by wire or wirelessly.

[0120] The information processing device 200 is a desktop personal computer, a notebook personal computer, a tablet terminal, or other information processing terminal.

[0121] In the example of FIG. 13, the learning unit 50b is implemented in the information processing device 200. Also, the determination unit 50c is implemented in the winding machine 1C.

[0122] The information processing device 200 collects the above-described learning data 123 (see FIG. 9) from the winding machines 1 (for example, winding machines 1A and 1B) connected to the network NW. Next, the learning unit 50b of the information processing device 200 executes a learning process using the learning data 123 collected from all the winding machines 1 and generates the above-described estimation model 124. Since the function of the learning unit 50b is as described above, the description thereof will not be repeated.

[0123] Thereafter, the information processing device 200 transmits the generated estimation model 124 to the winding machine 1 (for example, winding machine 1C). The determination unit 50c of the winding machine 1C uses the estimation model 124 to detect an abnormal mounting of the bobbin 91 mounted on the bobbin holders 41 and 42 of the winding machine 1. Since the function of the determination unit 50c is as described above, the description thereof will not be repeated.

[0124] <G. Second Modified Example> Next, a second modification of the above embodiment will be described with reference to FIGS.

[0125] The above-described mounting state detection device 10 identifies the bobbin range and the slit range in the judgment bobbin image, and detects a mounting abnormality of the bobbin 91 based on the bobbin range and the slit range. In contrast, in this modified example, the mounting state detection device 10 directly detects the type of mounting abnormality of the bobbin 91 from the judgment bobbin image. Note that other points such as the hardware configuration of the mounting state detection device 10 are as described above, and therefore will not be described repeatedly.

[0126] (G1. Training Dataset 122A) First, a description will be given of the learning data set 122A used for generating an estimation model that outputs the type of mounting abnormality of the bobbin 91 with reference to Fig. 14. Fig. 14 is a diagram showing an example of the learning data set 122A.

[0127] The training data set 122A includes a plurality of training data 123A. The number of training data 123A included in the training data set 122A is arbitrary. As an example, the number of training data 123A is several tens to several hundreds of thousands.

[0128] In each of the learning data 123A, mounting state information representing the mounting state of the bobbin 91 is associated as a label (correct value) with a learning bobbin image depicting the bobbin 91. In Fig. 14, the type of mounting abnormality of the bobbin 91 is shown as an example of the mounting state information.

[0129] The type of label defined in the learning data set 122A may be one type or may be multiple types. In the example of Fig. 14, as the labels of the mounting abnormality, an abnormality "A" indicating that the mounting orientation of the bobbin 91 is abnormal and an abnormality "B" indicating that the interval between the bobbins 91 is abnormal are shown.

[0130] (G2. Learning section 50e) Next, a learning unit 50e which is a modified example of the above-mentioned learning unit 50b (see FIG. 10) will be described with reference to Fig. 15. Fig. 15 is a diagram conceptually showing the learning process by the learning unit 50e in this modified example.

[0131] The learning unit 50e executes a learning process using the above-mentioned learning data set 122A (see FIG. 14) to generate an estimation model 124A for determining an abnormal attachment of the bobbin 91. The machine learning algorithm used in the learning process is not particularly limited, and may be, for example, deep learning, a convolutional neural network (CNN), a full-layer convolutional neural network (FCN), a support vector machine, or the like. The learning process using deep learning will be described below.

[0132] As shown in FIG. 15, the estimation model 124A is composed of an input layer X, an intermediate layer H, and an output layer Y.

[0133] The input layer X is configured to receive an input of a bobbin image of the learning data 123A. The input layer X includes, for example, N units x1 to x N (N is a natural number). The number of units that make up the input layer X is the same as the number of dimensions of the input information.

[0134] As an example, if the number of pixels of the bobbin image is N pixels and each pixel of the bobbin image is input directly to the input layer X, the input layer X is composed of N units. As another example, a feature extracted from the bobbin image may be input to the input layer X. In this case, the input layer X is configured so that the number of units is the same as the number of dimensions of the feature after feature extraction. Each unit constituting the input layer X outputs the input data to each unit in the first layer of the intermediate layer H.

[0135] The middle layer H is composed of one layer or multiple layers. In the example of FIG. 15, the middle layer H is composed of L layers (L is a natural number). Each layer of the middle layer H includes multiple units. The number of units in each layer of the middle layer H may be the same or different. In the example of FIG. 15, the first layer of the middle layer H is composed of Q units h A1 ~h AQ (Q is a natural number). The final layer of the hidden layer H is composed of R units h L1 ~h LR (R is a natural number).

[0136] Each unit constituting each layer of the intermediate layer H is connected to each unit in the previous layer and each unit in the next layer. Each unit in each layer receives an output value from each unit in the previous layer, multiplies each output value by a weight, accumulates the multiplication results, adds (or subtracts) a predetermined bias to (from) the accumulated result, inputs the addition result (or subtraction result) to a predetermined function (e.g., the Sigmonite function), and outputs the output value of the function to each unit in the next layer.

[0137] The output layer Y outputs an estimation result according to the input bobbin image. The output layer Y is composed of units y1 to y2, for example. Hereinafter, the units y1 to y2 are also referred to as units y.

[0138] Each unit y is connected to a unit h L1 ~h LR Each of the units y receives an output value from each unit in the final layer of the hidden layer H, multiplies each output value by a weight, accumulates the results of these multiplications, adds (or subtracts) a predetermined bias to (from) the accumulated result, inputs the result of the addition (or subtraction) to a predetermined function (e.g., the Sigmonite function), and outputs the output result of the function as an output value.

[0139] The number of units constituting the output layer Y is determined according to the number of types of wearing abnormalities defined in the learning data 123A. As an example, when detecting wearing abnormalities "A" and "B", the number of units constituting the output layer Y is two, units y1 and y2. In this case, unit y1 is configured to output a score "sa" indicating the possibility that wearing abnormality "A" has occurred. Unit y2 is configured to output a score "sb" indicating the possibility that wearing abnormality "B" has occurred.

[0140] 15, the estimation model 124A is configured to output multiple scores, but one estimation model may be configured to output one score. As an example, the first estimation model may be configured to output a score "sa" related to the anomaly type "A", and the second estimation model may be configured to output a score "sb" related to the anomaly type "B".

[0141] Next, the process of updating the internal parameters of the estimation model 124A by the learning unit 50e will be described.

[0142] The learning unit 50e inputs the bobbin image defined in the first learning data 123A to the estimation model 124A. Next, the learning unit 50e compares the estimation results “sa”, “sb” output from the estimation model 124A with the correct scores “sa′”, “sb′” according to the anomaly types associated with the first learning data 123A.

[0143] As an example, when the abnormality type associated with the learning data 123A is "A", the correct score is (sa', sb') = (1, 0). When the abnormality type associated with the learning data 123A is "B", the correct score is (sa', sb') = (0, 1).

[0144] The learning unit 50e calculates an error "Z" between the output results "sa", "sb" of the estimation model 124A and the correct scores "sa'", "sb'". The error "Z" is calculated, for example, based on the following formula (1).

[0145] Z={(sa-sa') 2 +(sb-sb') 2} / 2···(1) Next, the learning unit 50e updates various parameters (for example, weights and biases) included in the estimation model 124A so as to reduce the error "Z." The updating of the parameters is realized, for example, by the error backpropagation method.

[0146] The learning unit 50e repeatedly performs the update process of the internal parameters of the estimation model 124A for each piece of learning data 123A included in the learning data set 122A. As a result, the estimation model 124A starts to output accurate estimation results as the learning progresses.

[0147] The learning unit 50e does not need to use all of the learning data 123A included in the learning data set 122A for the learning process, and may generate the estimation model 124A using a portion of the learning data 123A included in the learning data set 122A. The remaining learning data 123A is used, for example, for evaluation of the estimation model 124A.

[0148] (G3. Judgment section 50f) Next, a function of a determination unit 50f which is a modification of the above-mentioned determination unit 50c (see FIG. 10) will be described with reference to Fig. 16. Fig. 16 is a diagram conceptually showing a process of determining whether or not the device is worn abnormally by the determination unit 50f according to this modification.

[0149] First, the determination unit 50f acquires the estimation model 124A generated by the learning unit 50e from a storage location. The acquisition location of the estimation model 124A may be the above-mentioned storage unit 50d or an external device.

[0150] Next, the determination unit 50f acquires a determination bobbin image from the imaging device 61 by imaging the bobbins 91 mounted on the bobbin holders 41 and 42. After that, the determination unit 50f inputs the determination bobbin image into the estimation model 124A. As a result, the estimation model 124A outputs a score indicating the possibility of a mounting abnormality of the bobbin 91 for each abnormality type. For example, as an estimation result, the estimation model 124A outputs a score "sa" indicating the possibility of the occurrence of a mounting abnormality "A" and a score "sb" indicating the possibility of the occurrence of a mounting abnormality "B".

[0151] Based on the scores "sa" and "sb", the determination unit 50f determines the type of the mounting abnormality that has occurred. As an example, when the score "sa" exceeds the first threshold value, the determination unit 50f determines that there is an abnormality in the mounting orientation of the bobbin 91. The first threshold value may be set in advance or may be arbitrarily set by the user.

[0152] Also, when the score "sb" exceeds the second threshold value, the determination unit 50f determines that there is an abnormality in the interval of the bobbins 91. The second threshold value may be set in advance or may be arbitrarily set by the user. Also, the second threshold value may be the same as the first threshold value or may be different from the first threshold value.

[0153] Further, when the score "sa" is less than or equal to the first threshold value and the score "sb" is less than or equal to the second threshold value, the determination unit 50f determines that the bobbin 91 is normally mounted on the bobbin holders 41 and 42.

[0154] Note that the output mode of the determination result by the determination unit 50f is not particularly limited. As an example, the determination unit 50f outputs a warning including the type of the mounting abnormality that has occurred. The warning may be displayed as a message on the above-described display device 106, may be output as a sound by the above-described notification unit 65, or may be stored in the storage unit 50d as a log.

[0155] <H. Modification Example 4> Next, referring to FIG. 17, a fourth modification of the above-described embodiment will be described. FIG. 17 is a side view of the bobbin winder 1 of the fourth modification.

[0156] In the above-described embodiment, one imaging device 61 images two bobbins 91. In contrast, in the fourth modification, one imaging device 61 images all the bobbins 91 mounted on the second bobbin holder 42. The imaging device 61 of the fourth modification is a wide-angle camera and can image a wide range. Also, in order to image all the bobbins 91, it is arranged at a certain distance from the bobbins 91. For example, the imaging device 61 of the fourth modification may be arranged at a position away from the bobbin winder 1.

[0157] <I. Modification 5> Next, referring to FIG. 18, a fifth modification of the above-described embodiment will be described. FIG. 18 is a side view of the bobbin winder 1 of the fifth modification.

[0158] In the above-described embodiment, the position of the imaging device 61 is fixed. In contrast, in the fifth modification, the imaging device 61 can travel along the axial direction of the second bobbin holder 42. Specifically, a rail 62 and a carriage 63 are provided. The rail 62 is formed along the axial direction of the second bobbin holder 42. The carriage 63 can travel along the rail 62. The imaging device 61 is arranged on the carriage 63. With this configuration, the imaging device 61 can be made to travel along the axial direction of the second bobbin holder 42.

[0159] By moving the imaging device 61, a large number of bobbins 91 can be imaged using one imaging device 61. The imaging device 61 may image the bobbins 91 while moving, or may image the bobbins 91 after moving and stopping at a specified position. Also, although one carriage 63 is shown in FIG. 18, a plurality of carriages 63 may be provided.

[0160] <J. Modification 6> Next, referring to FIG. 19, a sixth modification of the above-described embodiment will be described. FIG. 19 is a side view of the bobbin winder 1 of the sixth modification.

[0161] In the above embodiment, the wearing state detection device 10 detects whether the mounting state of the bobbins 91 is appropriate when all the bobbins 91 are mounted on the second bobbin holder 42. In contrast, in the sixth modification, the mounting state of the bobbin 91 is detected while the operation of inserting the bobbin 91 into the second bobbin holder 42 is being performed.

[0162] Specifically, in the sixth modification, the imaging device 61 is disposed near the insertion end of the second bobbin holder 42. The insertion end is the end on the side where the second bobbin holder 42 is inserted, in other words, the end on the side opposite to the turret plate 40. The bobbin 91 inserted from the insertion end is slid, and then the next bobbin 91 is inserted.

[0163] The imaging device 61 is disposed at a position where it can image the bobbin 91 inserted into the insertion end. The imaging device 61 images the bobbin 91 when the bobbin 91 is inserted. All the bobbins 91 inserted into the second bobbin holder 42 pass through the insertion end at least at the time of insertion. Also, the mounting orientation of the bobbin 91 does not change after insertion. Therefore, by imaging the bobbin 91 at the time of insertion and determining by the method of the above embodiment, it is possible to detect whether the insertion orientation of all the bobbins 91 is appropriate.

[0164] In the fourth to sixth modifications, the number of imaging devices 61 can be reduced, so there is a possibility of realizing low cost and space saving as compared with the above embodiment.

[0165] <K. Modification 7> Next, with reference to FIG. 20, a seventh modification of the above embodiment will be described. FIG. 20 is a front view of the winding machine 1 of the seventh modification.

[0166] In the above embodiment, the wearing state detection device 10 generates a bobbin image by imaging the bobbin 91 from one direction. Instead of this, in the seventh modification, the bobbin 91 is imaged from two directions to generate a bobbin image. Therefore, in the seventh modification, the imaging devices 61 are respectively disposed in different directions with respect to the bobbin 91.

[0167] In the seventh modification, since bobbin images of various locations on the outer circumferential surface of the bobbin 91 can be created, the mounting state of the bobbin 91 can be detected with higher accuracy. In particular, even if the slits 91a are not formed around the entire circumference of the bobbin 91, the slits 91a can be detected with higher accuracy.

[0168] The seventh modified example is applicable to any of the above-mentioned embodiment and the fourth to sixth modified examples. Moreover, the imaging device 61 may be moved to capture images of the bobbin 91 from two directions.

[0169] As described above, the mounting state detection device 10 of the above embodiment and modified example detects the mounting state of the bobbin 91 to the bobbin holder 41, 42. The mounting state detection device 10 includes the photographing device 61, the learning unit 50b, the storage unit 50d, and the determination unit 50c. The photographing device 61 photographs an area including the bobbin 91 to generate a bobbin image. The learning unit 50b constructs an estimation model by performing machine learning using the bobbin image and the area of ​​the bobbin 91 in the bobbin image as learning data. The storage unit 50d stores the estimation model constructed by the learning unit 50b. The determination unit 50c generates the area of ​​the bobbin 91 based on the bobbin image generated by the photographing device 61 and the estimation model stored in the storage unit 50d, and determines the mounting state of the bobbin 91 based on the area of ​​the bobbin 91.

[0170] By using machine learning, subtle characteristics and trends regarding the mounting state of the bobbin 91 can be taken into account, making it possible to accurately determine whether the bobbin 91 is properly mounted.

[0171] In the mounting state detection device 10 of the above embodiment and modified example, the bobbin 91 is formed with a slit 91a. The storage unit 50d stores the correct mounting orientation of the bobbin 91. The learning unit 50b constructs an estimation model by performing machine learning using the bobbin image and the range of the bobbin 91 and the range of the slit 91a in the bobbin image as learning data. The determination unit 50c generates the range of the bobbin 91 and the range of the slit 91a based on the bobbin image and the estimation model stored in the storage unit 50d. The determination unit 50c determines whether the mounting orientation of the bobbin 91 is correct or not based on the range of the bobbin 91 and the range of the slit 91a and the correct mounting orientation of the bobbin 91 stored in the storage unit 50d.

[0172] By using machine learning, the slit 91a can be accurately distinguished from other similar objects, and the mounting orientation of the bobbin 91 can be accurately determined.

[0173] In the attachment state detection device 10 of the above embodiment and modified example, a plurality of bobbins 91 are arranged side by side in the bobbin holders 41, 42. The memory unit 50d stores an appropriate range for the gap between adjacent bobbins 91. The determination unit 50c determines whether the gap between adjacent bobbins 91 is within an appropriate range based on the range of the plurality of bobbins 91 generated by the determination unit 50c and the appropriate range for the gap between adjacent bobbins 91 stored in the memory unit 50d.

[0174] By using machine learning, it is possible to accurately distinguish between the gaps between adjacent bobbins 91 and other similar objects, and therefore it is possible to accurately determine whether the gaps between the bobbins 91 are within an appropriate range.

[0175] The attachment state detection device 10 of the above embodiment and modified example includes a notification unit 65 that issues a notification when the determination unit 50c determines that the attachment state of the bobbin 91 is abnormal.

[0176] This makes it possible to notify the operator of any abnormality in the bobbin mounting state.

[0177] In the attachment state detection device 10 of the above embodiment and modified example, the photographing device 61 photographs an area including the bobbin 91 after the start of the work of attaching the bobbin 91 to the bobbin holders 41, 42 and before the yarn winding machine 1 starts winding the yarn 93.

[0178] This makes it possible to determine the bobbin installation state before a problem occurs due to incorrect bobbin installation.

[0179] In the mounting state detection device 10 of the above embodiment, a plurality of bobbins 91 are arranged in the bobbin holders 41 and 42. A plurality of photographing devices 61 are arranged along the axial direction of the second bobbin holder 42. The photographing device 61 generates a bobbin image including one or a plurality of bobbins 91.

[0180] As a result, even when a plurality of bobbins 91 are arranged side by side in the bobbin holders 41 and 42, the mounting state of these bobbins 91 can be detected with high accuracy by using a plurality of image capturing devices.

[0181] In the attachment state detection device 10 of the fourth modified example, a plurality of bobbins 91 are arranged in the bobbin holders 41 and 42. One image capturing device 61 generates a bobbin image including all the bobbins 91.

[0182] As a result, even when multiple bobbins 91 are arranged side by side in the bobbin holders 41, 42, by using a wide-angle camera 61 capable of photographing multiple bobbins 91 at once, the number of camera devices 61 can be reduced while the installation state of these bobbins 91 can be accurately detected.

[0183] In the attachment state detection device 10 of the fifth modified example, a plurality of bobbins 91 are arranged in line in the bobbin holders 41 and 42. The photographing device 61 is capable of traveling along the axial direction of the second bobbin holder 42. After generating a bobbin image including a bobbin 91, the photographing device 61 travels to generate a bobbin image including another bobbin 91.

[0184] As a result, even when multiple bobbins 91 are arranged side by side in the bobbin holders 41, 42, the number of photographing devices 61 can be reduced by running the photographing device 61 along the axial direction of the second bobbin holder 42, and the installation status of these bobbins 91 can be accurately detected.

[0185] In the attachment state detection device 10 of the sixth modified example, the bobbins 91 are inserted from the insertion ends of the bobbin holders 41, 42 and slid to arrange a plurality of bobbins 91 side by side. The photographing device 61 sequentially photographs the bobbins 91 inserted from the insertion ends to generate bobbin images for all of the bobbins 91.

[0186] As a result, even when multiple bobbins 91 are arranged side by side in the bobbin holders 41, 42, the installation status of these bobbins 91 can be accurately detected while reducing the number of photographing devices 61 by sequentially photographing the bobbins 91 as they are inserted.

[0187] The mounting state detection device 10 of the seventh modified example includes a plurality of photographing devices 61. The photographing devices 61 photograph the bobbin 91 from a plurality of directions. The determination unit 50c determines the mounting state of the bobbin 91 based on the plurality of bobbin images photographed from the plurality of directions.

[0188] This allows the determination to be made based on the results of photographing a wide range in the circumferential direction of the bobbin 91, thereby improving the detection accuracy.

[0189] In the mounting state detection device 10 of this embodiment, the photographing device 61 photographs the bobbin 91 multiple times while the bobbin 91 rotates around the axial direction of the bobbin holders 41, 42 as the center of rotation, to generate multiple photographing devices 61. The determination unit 50c determines the mounting state of one bobbin 91 based on the multiple bobbin images for that bobbin 91.

[0190] This allows the determination to be made based on the results of photographing the entire circumferential direction of the bobbin 91, thereby improving the detection accuracy.

[0191] Although the preferred embodiments of the present disclosure have been described above, the above configuration can be modified as follows, for example.

[0192] The flowchart shown in the above embodiment is an example, and some processes may be omitted, the content of some processes may be changed, or new processes may be added.

[0193] The traversing device 21 in the above embodiment is of a cam drum type, but it may have a different configuration as long as it is possible to reciprocate the traversing guide 23 in the width direction of the web. For example, instead of the traversing device 21, a belt type traversing device can also be used.

[0194] <L. Appendix> As described above, the present embodiment includes the following disclosure.

[0195] From the viewpoint of the present disclosure, a mounting state detection device having the following configuration is provided. That is, the mounting state detection device detects the mounting state of the bobbin on the bobbin holder of the winding machine. The mounting state detection device includes a photographing device, a learning unit, a storage unit, and a determination unit. The photographing device photographs the range including the bobbin to generate a bobbin image. The learning unit constructs an estimation model by performing machine learning using the bobbin image and the range of the bobbin in the bobbin image as learning data. The storage unit stores the estimation model constructed by the learning unit. The determination unit generates the range of the bobbin based on the bobbin image generated by the photographing device and the estimation model stored in the storage unit, and determines the mounting state of the bobbin based on the range of the bobbin.

[0196] By using machine learning, it is possible to consider slight features and trends regarding the mounting state of the bobbin, so it is possible to accurately determine whether the bobbin is properly mounted.

[0197] The above-mentioned mounting state detection device is preferably configured as follows. That is, a slit is formed in the bobbin. The storage unit stores the correct mounting orientation of the bobbin. The learning unit constructs an estimation model by performing machine learning using the bobbin image and the range of the bobbin and the range of the slit in the bobbin image as learning data. The determination unit generates the range of the bobbin and the range of the slit based on the bobbin image and the estimation model stored in the storage unit. The determination unit determines whether the mounting orientation of the bobbin is correct or not based on the range of the bobbin and the range of the slit, and the correct mounting orientation of the bobbin stored in the storage unit.

[0198] By using machine learning, it is possible to accurately distinguish between slits and other similar objects, thereby enabling the installation orientation of the bobbin to be determined with high accuracy.

[0199] The mounting state detection device is preferably configured as follows: That is, a plurality of the bobbins are arranged in the bobbin holder. The memory unit stores an appropriate range of the gap between adjacent bobbins. It is preferable that the determination unit determines whether the gap between adjacent bobbins is within an appropriate range based on the range of the plurality of bobbins generated by the determination unit and the appropriate range of the gap between adjacent bobbins stored in the memory unit.

[0200] By using machine learning, it is possible to accurately distinguish between gaps between adjacent bobbins and other similar objects, making it possible to accurately determine whether the bobbin gaps are within an appropriate range.

[0201] The mounting state detection device preferably further includes a notification unit that issues a notification when the determination unit determines that the mounting state of the bobbin is abnormal.

[0202] This makes it possible to notify the operator of any abnormality in the bobbin mounting state.

[0203] In the above-mentioned attachment state detection device, it is preferable that the photographing device photographs an area including the bobbin after the start of the work of attaching the bobbin to the bobbin holder and before the yarn winding machine starts winding the yarn.

[0204] This makes it possible to determine the bobbin installation state before a problem occurs due to incorrect bobbin installation.

[0205] The mounting state detection device is preferably configured as follows: A plurality of the bobbins are arranged in the bobbin holder. A plurality of the photographing devices are arranged along the axial direction of the bobbin holder. The photographing devices generate the bobbin image including one or a plurality of the bobbins.

[0206] As a result, even when a plurality of bobbins are arranged side by side in the bobbin holder, the mounting state of these bobbins can be detected with high accuracy by using a plurality of image capturing devices.

[0207] The above-mentioned mounting state detection device is preferably configured as follows: A plurality of the bobbins are arranged in the bobbin holder, and one of the photographing devices generates the bobbin image including all of the bobbins.

[0208] As a result, even when multiple bobbins are arranged side by side in a bobbin holder, by using a wide-angle camera that can photograph multiple bobbins at once, the number of camera devices can be reduced while the installation status of these bobbins can be accurately detected.

[0209] The above-mentioned mounting state detection device is preferably configured as follows: a plurality of the bobbins are arranged in the bobbin holder; the photographing device is movable along the axial direction of the bobbin holder; and after generating the bobbin image including the bobbin, the photographing device moves to generate the bobbin image including another bobbin.

[0210] As a result, even when multiple bobbins are arranged side by side in the bobbin holder, the installation status of these bobbins can be accurately detected while reducing the number of photographing devices by running the photographing device along the axial direction of the bobbin holder.

[0211] The mounting state detection device is preferably configured as follows: That is, the bobbins are inserted from an insertion end of the bobbin holder and slid to arrange a plurality of the bobbins side by side, and the photographing device sequentially photographs the bobbins inserted from the insertion end to generate the bobbin images for all of the bobbins.

[0212] As a result, even when multiple bobbins are arranged side by side in the bobbin holder, the installation status of these bobbins can be accurately detected while reducing the number of photographing devices by sequentially photographing the bobbins as they are inserted.

[0213] The mounting state detection device is preferably configured as follows: That is, the mounting state detection device includes a plurality of the photographing devices. The photographing devices photograph the bobbin from a plurality of directions. The determination unit determines the mounting state of the bobbin based on the plurality of bobbin images photographed from the plurality of directions.

[0214] This allows the determination to be made based on the results of photographing a wide range in the circumferential direction of the bobbin, thereby improving the detection accuracy.

[0215] The mounting state detection device is preferably configured as follows: That is, the photographing device photographs the bobbin a plurality of times while the bobbin is rotating about an axial direction of the bobbin holder as a rotation center to generate a plurality of bobbin images, and the determining unit determines the mounting state of the bobbin based on the plurality of bobbin images for one of the bobbins.

[0216] This allows the determination to be made based on the results of photographing the entire circumferential direction of the bobbin, thereby improving the detection accuracy.

[0217] The embodiments disclosed herein should be considered to be illustrative and not restrictive in all respects. The scope of the present invention is defined by the claims, not the above description, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0218] 1 Yarn winding machine 10. Wearing state detection device 50 Control device 50a Control section 50b Learning Section 50c Judgment part 61 Imaging Device 65 Notification Department 124 Estimation Model

Claims

1. Control device and The system includes a photographic device configured to photograph bobbins mounted on the bobbin holder of a thread winding machine, The control device is The process involves obtaining an estimation model generated by machine learning using multiple training data sets, and each of the multiple training data sets associates a training bobbin image of a bobbin with mounting status information representing the mounting state of the bobbin as a label. The control device further, The process involves obtaining a judgment bobbin image from the imaging device by photographing the bobbin mounted on the bobbin holder, The process involves inputting the aforementioned bobbin image for determination into the estimation model, and then determining whether or not the bobbin is properly mounted to the bobbin holder based on the mounting status information output from the estimation model. In the process described above, it is determined whether the orientation of the bobbin mounted on the bobbin holder is correct. The bobbin has a slit formed in it. The mounting status information, which serves as the label, includes a bobbin range indicating the extent of the bobbin within the learning bobbin image and a slit range indicating the extent of the slit within the learning bobbin image. The estimation model receives the judgment bobbin image as input and outputs the bobbin range and the slit range within the judgment bobbin image as mounting state information. A mounting state detection device that determines whether the mounting orientation is normal based on the positional relationship between the bobbin range and the slit range output from the estimation model.

2. A control device, The system includes a photographic device configured to photograph bobbins mounted on the bobbin holder of a thread winding machine, The control device is The process involves obtaining an estimation model generated by machine learning using multiple training data sets, and each of the multiple training data sets associates a training bobbin image of a bobbin with mounting status information representing the mounting state of the bobbin as a label. The control device further, The process involves obtaining a judgment bobbin image from the imaging device by photographing the bobbin mounted on the bobbin holder, The process involves inputting the aforementioned bobbin image for determination into the estimation model, and then determining whether or not the bobbin is properly mounted to the bobbin holder based on the mounting status information output from the estimation model. Multiple bobbins are arranged side by side in the bobbin holder. The mounting state detection device determines whether the spacing between each bobbin included in the determination bobbin image is normal in the determination process.

3. The mounting status information, which serves as the label, includes a bobbin range indicating the extent of the bobbin within the learning bobbin image. The estimation model receives the judgment bobbin image as input and outputs the bobbin range for each bobbin included in the judgment bobbin image. The mounting state detection device according to claim 2, wherein whether the aforementioned interval is normal is determined based on the positional relationship between adjacent bobbin ranges within each bobbin range output from the estimation model.

4. The mounting status detection device according to any one of claims 1 to 3, further comprising: the control device further executing a process to notify of a mounting abnormality of the bobbin when it is determined that the bobbin is not properly mounted to the bobbin holder.

5. The mounting state detection device according to any one of claims 1 to 3, wherein the photographing device photographs the bobbin after the start of the work of mounting the bobbin to the bobbin holder and before the start of winding the thread by the thread winding machine.

6. The mounting state detection device according to claim 4, wherein the photographing device photographs the bobbin after the start of the work of mounting the bobbin in the bobbin holder and before the start of winding the thread by the thread winding machine.

7. Multiple bobbins are arranged in a row on the bobbin holder. Multiple imaging devices are arranged along the axial direction of the bobbin holder. The mounting state detection device according to any one of claims 1 to 3, wherein the imaging device generates the determination bobbin image including one or more bobbins.

8. The bobbins are arranged in a row on the bobbin holder, Multiple imaging devices are arranged along the axial direction of the bobbin holder. The mounting state detection device according to claim 4, wherein the imaging device generates the determination bobbin image including one or more of the bobbins.

9. The bobbins are arranged in a row on the bobbin holder, Multiple imaging devices are arranged along the axial direction of the bobbin holder. The mounting state detection device according to claim 5, wherein the imaging device generates the determination bobbin image including one or more of the bobbins.

10. The bobbins are arranged in a row on the bobbin holder, Multiple imaging devices are arranged along the axial direction of the bobbin holder. The mounting state detection device according to claim 6, wherein the imaging device generates the determination bobbin image including one or more of the bobbins.

11. Multiple bobbins are arranged in a row on the bobbin holder. The mounting state detection device according to any one of claims 1 to 3, wherein one of the imaging devices generates the determination bobbin image including all of the bobbins.

12. The bobbins are arranged in a row on the bobbin holder, The mounting state detection device according to claim 4, wherein one of the imaging devices generates the determination bobbin image including all of the bobbins.

13. The bobbins are arranged in a row on the bobbin holder, The mounting state detection device according to claim 5, wherein one of the imaging devices generates the determination bobbin image including all of the bobbins.

14. The bobbins are arranged in a row on the bobbin holder, The mounting state detection device according to claim 6, wherein one of the imaging devices generates the determination bobbin image including all of the bobbins.

15. Multiple bobbins are arranged in a row on the bobbin holder. The aforementioned imaging device is configured to be movable along the axial direction of the bobbin holder. The mounting state detection device according to any one of claims 1 to 3, wherein the imaging device generates the determination bobbin image including the bobbin, then travels to generate another determination bobbin image including the bobbin.

16. The bobbins are arranged in a row on the bobbin holder, The aforementioned imaging device is configured to be movable along the axial direction of the bobbin holder. The mounting state detection device according to claim 4, wherein the imaging device generates the determination bobbin image including the bobbin, then travels to generate another determination bobbin image including the bobbin.

17. The bobbins are arranged in a row on the bobbin holder, The aforementioned imaging device is configured to be movable along the axial direction of the bobbin holder. The mounting state detection device according to claim 5, wherein the imaging device generates the determination bobbin image including the bobbin, then travels to generate another determination bobbin image including the bobbin.

18. The bobbins are arranged in a row on the bobbin holder, The aforementioned imaging device is configured to be movable along the axial direction of the bobbin holder. The mounting state detection device according to claim 6, wherein the imaging device generates the determination bobbin image including the bobbin, then travels to generate another determination bobbin image including the bobbin.

19. The bobbin is inserted from the insertion end of the bobbin holder and the bobbin slides, so that multiple bobbins are arranged in a row. The mounting state detection device according to any one of claims 1 to 3, wherein the imaging device generates the determination bobbin image for all of the bobbins by sequentially photographing the bobbins inserted from the insertion end.

20. The bobbin is inserted from the insertion end of the bobbin holder and the bobbin slides, so that multiple bobbins are arranged in a row, The mounting state detection device according to claim 4, wherein the imaging device generates the determination bobbin image for all of the bobbins by sequentially photographing the bobbins inserted from the insertion end.

21. The bobbin is inserted from the insertion end of the bobbin holder and the bobbin slides, so that multiple bobbins are arranged in a row, The mounting state detection device according to claim 5, wherein the imaging device generates the determination bobbin image for all of the bobbins by sequentially photographing the bobbins inserted from the insertion end.

22. The bobbin is inserted from the insertion end of the bobbin holder and the bobbin slides, so that multiple bobbins are arranged in a row, The mounting state detection device according to claim 6, wherein the imaging device generates the determination bobbin image for all of the bobbins by sequentially photographing the bobbins inserted from the insertion end.

23. The camera is equipped with multiple such devices. The aforementioned imaging device photographs the bobbin from multiple directions, The mounting state detection device according to any one of claims 1 to 3, wherein whether or not a bobbin is properly mounted in the bobbin holder is determined based on a plurality of determination bobbin images taken from multiple directions.

24. comprising a plurality of the above-mentioned imaging devices, The aforementioned imaging device photographs the bobbin from multiple directions, The mounting state detection device according to claim 4, wherein whether or not a bobbin is properly mounted in the bobbin holder is determined based on a plurality of determination bobbin images taken from multiple directions.

25. comprising a plurality of the above-mentioned imaging devices, The aforementioned imaging device photographs the bobbin from multiple directions, The mounting state detection device according to claim 5, wherein whether or not a bobbin is properly mounted in the bobbin holder is determined based on a plurality of determination bobbin images taken from multiple directions.

26. comprising a plurality of the above-mentioned imaging devices, The aforementioned imaging device photographs the bobbin from multiple directions, The mounting state detection device according to claim 6, wherein whether or not a bobbin is properly mounted in the bobbin holder is determined based on a plurality of determination bobbin images taken from multiple directions.

27. The imaging device generates multiple images of the bobbin for determination by taking multiple images of the bobbin while the bobbin is rotating with the axial direction of the bobbin holder as the center of rotation. The mounting state detection device according to any one of claims 1 to 3, wherein whether or not a bobbin is properly mounted in the bobbin holder is determined based on a plurality of determination bobbin images.

28. The imaging device generates multiple images of the bobbin for determination by taking multiple images of the bobbin while the bobbin is rotating with the axial direction of the bobbin holder as the center of rotation, The mounting state detection device according to claim 4, wherein whether or not a bobbin is properly mounted in the bobbin holder is determined based on a plurality of determination bobbin images.

29. The imaging device generates multiple images of the bobbin for determination by taking multiple images of the bobbin while the bobbin is rotating with the axial direction of the bobbin holder as the center of rotation, The mounting state detection device according to claim 5, wherein whether or not a bobbin is properly mounted to the bobbin holder is determined based on a plurality of determination bobbin images.

30. The imaging device generates multiple images of the bobbin for determination by taking multiple images of the bobbin while the bobbin is rotating with the axial direction of the bobbin holder as the center of rotation, The mounting state detection device according to claim 6, wherein whether or not a bobbin is properly mounted in the bobbin holder is determined based on a plurality of determination bobbin images.

31. A method for detecting the mounting state, which is performed by a mounting state detection device, The mounting state detection device includes a camera configured to photograph the bobbin mounted on the bobbin holder of the thread winding machine, The aforementioned mounting state detection method is: The process includes a step of obtaining an estimation model generated by machine learning using multiple training data sets, wherein each of the multiple training data sets associates a training bobbin image of a bobbin with mounting status information representing the mounting state of the bobbin as a label. The aforementioned mounting state detection method further includes: The steps include: acquiring a bobbin image for determination from the imaging device by photographing the bobbin mounted on the bobbin holder; The system includes the step of inputting the determination bobbin image into the estimation model and determining whether or not the bobbin is properly mounted to the bobbin holder based on the mounting status information output from the estimation model, In the above determination step, it is determined whether the orientation of the bobbin mounted on the bobbin holder is correct. The bobbin has a slit formed in it. The mounting status information, which serves as the label, includes a bobbin range indicating the extent of the bobbin within the learning bobbin image and a slit range indicating the extent of the slit within the learning bobbin image. The estimation model receives the judgment bobbin image as input and outputs the bobbin range and the slit range within the judgment bobbin image as mounting state information. A mounting state detection method in which whether or not the mounting orientation is normal is determined based on the positional relationship between the bobbin range and the slit range output from the estimation model.

32. A method for detecting the mounting state, which is performed by a mounting state detection device, The mounting state detection device includes a camera configured to photograph the bobbin mounted on the bobbin holder of the thread winding machine, The aforementioned mounting state detection method is: The process includes a step of obtaining an estimation model generated by machine learning using multiple training data sets, wherein each of the multiple training data sets associates a training bobbin image of a bobbin with mounting status information representing the mounting state of the bobbin as a label. The aforementioned mounting state detection method further includes: The steps include: acquiring a bobbin image for determination from the imaging device by photographing the bobbin mounted on the bobbin holder; The system includes the step of inputting the determination bobbin image into the estimation model and determining whether or not the bobbin is properly mounted to the bobbin holder based on the mounting status information output from the estimation model, Multiple bobbins are arranged side by side in the bobbin holder. The mounting state detection method, in the determination step, determines whether the spacing between each bobbin included in the determination bobbin image is normal.

33. A mounting status detection program executed by a mounting status detection device, The mounting state detection device includes a camera configured to photograph the bobbin mounted on the bobbin holder of the thread winding machine, The mounting state detection program is provided to the mounting state detection device. The process involves obtaining an estimation model generated by machine learning using multiple training data sets, where each of the multiple training data sets associates a training bobbin image of a bobbin with mounting status information representing the mounting state of the bobbin as a label. The aforementioned mounting state detection program is further provided to the mounting state detection device, The steps include: acquiring a bobbin image for determination from the imaging device by photographing the bobbin mounted on the bobbin holder; The determination bobbin image is input to the estimation model, and the model is made to perform the step of determining whether or not the bobbin is properly mounted on the bobbin holder based on the mounting status information output from the estimation model. In the above determination step, it is determined whether the orientation of the bobbin mounted on the bobbin holder is correct. The bobbin has a slit formed in it. The mounting status information, which serves as the label, includes a bobbin range indicating the extent of the bobbin within the learning bobbin image and a slit range indicating the extent of the slit within the learning bobbin image. The estimation model receives the judgment bobbin image as input and outputs the bobbin range and the slit range within the judgment bobbin image as mounting state information. A mounting state detection program determines whether the mounting orientation is correct based on the positional relationship between the bobbin range and the slit range output from the estimation model.

34. A mounting state detection program executed by a mounting state detection device, The mounting state detection device includes a camera configured to photograph the bobbin mounted on the bobbin holder of the thread winding machine, The mounting state detection program is provided to the mounting state detection device. The process involves obtaining an estimation model generated by machine learning using multiple training data sets, where each of the multiple training data sets associates a training bobbin image of a bobbin with mounting status information representing the mounting state of the bobbin as a label. The aforementioned mounting state detection program is further provided to the mounting state detection device, The steps include: acquiring a bobbin image for determination from the imaging device by photographing the bobbin mounted on the bobbin holder; The determination bobbin image is input to the estimation model, and the model is made to perform the step of determining whether or not the bobbin is properly mounted on the bobbin holder based on the mounting status information output from the estimation model. Multiple bobbins are arranged side by side in the bobbin holder. The mounting state detection program, in the determination step, determines whether the spacing between each bobbin included in the determination bobbin image is normal.

Images