Abnormality determination device, abnormality determination method, and abnormality determination program

Abstract

This invention provides an anomaly determination device, an anomaly determination method, and an anomaly determination program capable of accurately determining anomalies in yarn. The control device includes an input unit, a determination unit, and an output unit. A thermal image of a yarn path region, which is part of the first yarn guide (5a) of the spinning traction machine (1), is input to the input unit. Based on the thermal image input to the input unit, the determination unit determines whether there is an anomaly related to the yarn (93). The output unit outputs the determination result of the determination unit.

Description

Abnormal determination device, abnormal determination method, and abnormal determination program The present invention mainly relates to an abnormal determination device that outputs a determination result of an abnormality related to a wire in a synthetic fiber manufacturing facility. Patent Document 1 discloses a defect detection device that detects spinning defects when multiple wires are spun into a shape for fabric manufacturing. The defect detection device includes a CCD (charge-coupled device) camera and a data processing unit. The CCD camera captures multiple wires and outputs an image signal to the data processing unit. The data processing unit detects spinning defects based on the image signal. Specifically, it is previously stored which position (which pixel) of the image signal corresponds to each of the multiple wires. For example, in the case of an abnormality such as a position shift or breakage of a wire, the detection intensity of the position (pixel) corresponding to the wire with the abnormality decreases. Therefore, the data processing unit detects spinning defects based on the detection intensity of each pixel of the image signal. In Patent Document 1, it is previously stored which pixel of the image signal corresponds to the wire. However, generally, since the width of the wire is thin and the position of the wire is likely to change, it may frequently occur that the wire is not displayed at the corresponding pixel. That is, by simply analyzing the visible image including the wire, it may be impossible to detect abnormalities related to the wire with high accuracy. Prior Art Documents Patent Documents Patent Document 1: Japanese Patent Application Laid-Open No. 11-279923 The present invention has been completed in view of the above circumstances, and its main object is to provide an abnormal determination device that can accurately determine an abnormality related to a wire. In one example of the present invention, an abnormal determination device is provided. The abnormal determination device includes: an input unit for inputting a thermal image including a wire path region, the wire path region being a part of a wire guide included in a synthetic fiber manufacturing facility; a determination unit for determining whether there is an abnormality related to a wire based on the thermal image input to the input unit; and an output unit for outputting the determination result of the determination unit. In one example of the present invention, the determination unit determines whether there is the above abnormality based on the determination thermal image input to the input unit and a determination model generated by machine learning on a plurality of learning data. In each of the plurality of learning data, the presence or absence of an abnormality related to a wire passing through a wire guide included in the synthetic fiber manufacturing facility or another synthetic fiber manufacturing facility is associated with a learning thermal image that maps the wire guide. In one example of the present invention, the abnormal determination device further includes a learning unit for generating the determination model based on the plurality of learning data. In one example of the present invention, the above abnormality includes an abnormality of the wire path. In one instance of the present invention, the above anomalies include anomalies in strand tension. In one instance of the present invention, the above anomaly includes an anomaly in the amount of adhesion of the oil agent imparted to the thread. In one instance of the present invention, the above-mentioned anomaly judgment device has a plurality of the above-mentioned judgment models obtained by learning separately according to similar operating conditions. The above judgment part selects the judgment model corresponding to the current operating condition from among the plurality of the above judgment models, and determines the presence or absence of the above anomalies based on the selected judgment model and the above thermal image used for the judgment input to the said input part. In one example of the present invention, the above-mentioned synthetic fiber manufacturing equipment is thereby equipped with a shooting device capable of photographing the said thread guide. Obtain the above thermal image input to the above input section from the above shooting device. In one instance of the invention, the above anomaly judgment device has a plurality of the above judgment models obtained by learning separately according to the positional relationship between the thread guide and the shooting device, which is provided in the above-mentioned synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment described above. The above judgment part selects the judgment model corresponding to the positional relationship between the said thread guide and the above shooting device equipped with the above mentioned synthetic fiber manufacturing equipment from among the several above judgment models, based on the selected judgment model and the above thermal image used for the judgment input to the said input section, to determine the presence of the above anomalies. In one example of the present invention, the synthetic fiber manufacturing equipment described above is a spinning traction machine. In one instance of the present invention, the above-mentioned synthetic fiber manufacturing equipment is a dummy machine. In other instances of the invention, an anomaly determination method is provided. The above anomaly determination method has: the step of inputting a thermal image including the thread channel region, which is part of the thread guide equipped with the synthetic fiber manufacturing equipment; In one instance of the invention, in the step of the above judgment, the presence or absence of the above anomalies is determined based on the above thermal image for the judgment entered in the step of the above input and the judgment model generated by resorting to machine learning of multiple learning data. In each of the above plurality of learning-use data, an association was established with the presence or absence of an abnormal thread traversing the thread guides equipped with the above-mentioned synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment as a label to the thermal imagery mapping the learning use of that thread guide. In one instance of the present invention, the above-mentioned anomaly judgment method further has the step of generating the above-mentioned judgment model based on the above-mentioned plurality of learning data. In one instance of the invention, the above anomaly includes an anomaly of a thread channel. In one example of the present invention, the above abnormal determination method further includes the step of obtaining a plurality of the above determination models learned respectively according to similar operating conditions. The above determination steps include: the step of selecting a determination model corresponding to the current operating condition from among the plurality of the above determination models; and the step of determining whether there is the above abnormality based on the selected determination model and the above thermal image for determination input in the above input step. In one example of the present invention, the above synthetic fiber manufacturing equipment further includes a photographing device configured to be able to photograph the above wire guide. The above thermal image input in the above input step is obtained from the above photographing device. In one example of the present invention, the above abnormal determination method further includes the step of obtaining a plurality of the above determination models learned respectively according to the positional relationship between the wire guide and the photographing device provided in the above synthetic fiber manufacturing equipment or the above other synthetic fiber manufacturing equipment. The above determination steps include: the step of selecting a determination model corresponding to the positional relationship between the wire guide and the photographing device provided in the above synthetic fiber manufacturing equipment from among the plurality of the above determination models; and the step of determining whether there is the above abnormality based on the selected determination model and the above thermal image for determination input in the above input step. In other examples of the present invention, there is provided an abnormal determination program. The above abnormal determination program causes an abnormal determination device to execute the following steps: the step of inputting a thermal image including a wire path region, the wire path region being a part of a wire guide provided in a synthetic fiber manufacturing equipment; the step of determining whether there is an abnormality related to a wire based on the above thermal image input in the above input step; and the step of outputting the determination result of the above determination step. In one example of the present invention, in the above determination step, it is determined whether there is the above abnormality based on the above thermal image for determination input in the above input step and a determination model generated by performing machine learning on a plurality of learning materials. In each of the plurality of learning materials, the presence or absence of an abnormality related to a wire passing through a wire guide provided in the above synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment is associated as a label with a learning thermal image mapping the wire guide. In one example of the present invention, the above abnormal determination program causes the above abnormal determination device to further execute the step of generating the above determination model according to the above plurality of learning materials. In one example of the present invention, the above abnormality includes an abnormality of the wire path. In one example of the present invention, the above abnormal determination program causes the above abnormal determination device to further execute the step of obtaining a plurality of the above determination models respectively learned according to similar operating conditions. The above determination steps include: a step of selecting a determination model corresponding to the current operating condition from among the plurality of the above determination models; and a step of determining whether there is the above abnormality based on the selected determination model and the above thermal image for determination input in the above input step. In one example of the present invention, the above synthetic fiber manufacturing equipment further includes a photographing device configured to be able to photograph the above godet. The above thermal image input in the above input step is obtained from the above photographing device. In one example of the present invention, the above abnormal determination program causes the above abnormal determination device to further execute the step of obtaining a plurality of the above determination models respectively learned according to the positional relationship between the godet and the photographing device provided in the above synthetic fiber manufacturing equipment or the above other synthetic fiber manufacturing equipment. The above determination steps include: a step of selecting a determination model corresponding to the positional relationship between the above godet and the above photographing device provided in the above synthetic fiber manufacturing equipment from among the plurality of the above determination models; and a step of determining whether there is the above abnormality based on the selected determination model and the above thermal image for determination input in the above input step. Hereinafter, each embodiment of the present invention will be described with reference to the accompanying drawings. In the following description, the same reference numerals are given to the same components and constituent elements. Their names and functions are also the same. Therefore, detailed descriptions thereof will not be repeated. It should be noted that the following-described embodiments and each modification example can also be appropriately and selectively combined. <A. Spinning and drawing machine 1> Next, an embodiment of the present invention will be described with reference to the accompanying drawings. FIG. 1 is a side view of a spinning and drawing machine (synthetic fiber manufacturing equipment) 1 according to an embodiment of the present invention. The spinning and drawing machine 1 shown in FIG. 1 includes a spinning device 2 and a yarn winding machine 9. The spinning device 2 includes a spinneret 2a. A molten fiber material such as nylon or polyester is supplied from a raw material supply device (not shown) to the spinning device 2. The spinning device 2 extrudes the molten fiber material in a high-temperature state from the spinneret 2a. Thereby, a plurality of yarns 93 are spun from the spinneret 2a. The yarn winding machine 9 winds the yarns 93 spun by the spinning device 2 to manufacture a package 94. It should be noted that the yarns 93 are, for example, synthetic fiber yarns such as nylon or polyester. The yarns 93 are long fiber yarns (filaments), not short fiber yarns such as textile yarns. In addition, the yarns 93 can also be elastic yarns such as elastic fibers (Spandex). In the traveling direction of the yarns 93, an oil supply guide 3, a yarn cutting and suction device 4, a first godet 5a, a first godet roller 7, a second godet 5b, a second godet roller 8, and a third godet 5c are configured between the spinning device 2 and the yarn winding machine 9. The oil supply guide 3 guides the filament 93 spun from the spinning device 2 and applies an oil agent to the filament 93. The filament cutting and suction device 4 is configured downstream of the oil supply guide 3. A drive device 4a and a suction source 4b are connected to the filament cutting and suction device 4. The filament cutting and suction device 4 can move forward and backward relative to the filament 93 by the driving force of the drive device 4a. The filament cutting and suction device 4 can suck the filament 93 by the negative pressure of the suction source 4b. The first wire guide 5a is configured downstream of the filament cutting and suction device 4. The first wire guide 5a guides multiple filaments 93 separately. The detailed shape of the first wire guide 5a will be described later. A photographing device 6 is provided around the first wire guide 5a. The photographing device 6 is configured with the lens facing the first wire guide 5a, photographs the first wire guide 5a, and generates an image. The photographing device 6 is a thermal imaging camera, measures the temperature of an object based on the intensity of far-infrared rays radiated from the object, and generates a thermal image. A thermal image refers to an image generated with different concentrations, brightness, chromaticity, etc. based on the temperature of an object. It should be noted that the thermal image can also be the raw data output from the thermal imaging camera, and does not necessarily have to be an image generated based on the intensity of far-infrared rays radiated from the object. This raw data represents, for example, the temperature of each part of the object in an array. In addition, the thermal image can also be obtained from a temperature sensor or the like, and does not necessarily have to be obtained from the photographing device 6. The photographing device 6 photographs the filament path area. The filament path area represents the area through which the filament path (the path through which the filament 93 passes) passes. Specifically, it is the area assumed to be passed through by the filament path under normal conditions, and the filament 93 may not pass through the filament path area under abnormal conditions. The filament path area is the path of the filament 93 assumed during design, and is derived, for example, based on the positions of various components guiding the filament 93. Specifically, the photographing device 6 photographs the filament path area including a part of the first wire guide 5a and generates a thermal image. The thermal image generated by the photographing device 6 is used for the determination of abnormalities related to the filament 93 as described later. The first wire guide roller 7 and the second wire guide roller 8 are driven by a motor (not shown). The first wire guide roller 7 and the second wire guide roller 8 can牵引 the filament 93 by winding the filament 93 and rotating. In addition, in the traveling direction of the filament 93, a second wire guide 5b is configured between the first wire guide roller 7 and the second wire guide roller 8. The second wire guide 5b guides multiple filaments 93 separately from the first wire guide roller 7 to the second wire guide roller 8. The first wire guide roller 7 is configured downstream of the first wire guide 5a. The first wire guide roller 7牵引 the filament 93 from the first wire guide 5a. The second wire guide roller 8 is configured downstream of the first wire guide roller 7. The second wire guide roller 8牵引 the filament 93 from the first wire guide roller 7. A filament winding machine 9 is configured downstream of the second wire guide roller 8. The filament 93 passing through the second wire guide roller 8 is guided by the third wire guide 5c and supplied to the filament winding machine 9. The third wire guide 5c is configured between the second wire guide roller 8 and a later-described traverse device 21 in the traveling direction of the filament 93. The third wire guide 5c guides multiple filaments 93 separately. Next, referring to FIGS. 1 to 3, the thread winding machine 9 will be described. As shown in FIGS. 1 and 2, the thread winding machine 9 includes a frame 11, a first housing 20, a second housing 30, and a turntable 40. The frame 11 is a member that holds each part of the thread winding machine 9. The first housing 20 and the second housing 30 are assembled to the frame 11. The first housing 20 and the second housing 30 can be raised and lowered relative to the frame 11. A traverse device 21 is assembled to the first housing 20. In a state where the traverse guide 23 described later is engaged with the thread 93, the traverse device 21 reciprocates in the axial direction of the first bobbin holder 41 described later over the winding width of the package 94, thereby causing the thread 93 fed to the downstream side to traverse. By this traversing action of the thread 93, a thread layer is formed 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. The traverse cam 22 is a roller-shaped member configured in parallel with the bobbin 91 or the package 94. A spiral cam groove is formed on the outer peripheral surface of the traverse cam 22. The traverse cam 22 is rotationally driven by a traverse motor 51. The traverse motor 51 is controlled by a control device 50 described later. The traverse guide 23 is a part that engages with the thread 93. The tip of the traverse guide 23 has, for example, a substantially U-shaped guide portion and engages with the thread 93 in a manner that sandwiches the thread 93 in the winding width direction. The base end of the traverse guide 23 is located in the cam groove of the traverse cam 22. According to this configuration, the traverse guide 23 can be reciprocated in the winding width direction by rotationally driving the traverse cam 22. A contact roller 31 is rotatably assembled to the second housing 30. The contact roller 31 is rotatably driven while contacting the thread layer of the package 94 with a predetermined pressure during the winding of the thread 93, thereby feeding the thread 93 from the traverse guide 23 to the thread layer of the package 94 and shaping the thread layer of the package 94. It should be noted that a drive unit such as a motor can also be used to rotationally drive the contact roller 31. An operation panel 32 is provided on the second housing 30. The operation panel 32 is a device operated by an operator. The operator gives instructions to the thread winding machine 9 by operating the operation panel 32. Examples of the instructions given by the operator include the start of winding, the stop of winding, and the change of winding conditions. The turntable 40 is a member having a disc shape. The turntable 40 is rotatably assembled to the frame 11. The turntable 40 can rotate about an axis passing through the center of the disc. The turntable 40 is rotationally driven by a turntable motor 53 as shown in FIG. 3. The turntable motor 53 is controlled by a control device 50 described later. At two positions in the turntable 40 that are opposed to each other across the center of the circular plate, a first bobbin holder 41 and a second bobbin holder 42 are respectively provided. A plurality of bobbins 91 can be attached to the first bobbin holder 41 in a manner arranged in the axial direction of the first bobbin holder 41. A plurality of bobbins 91 can be attached to the second bobbin holder 42 in a manner arranged in the axial direction of the second bobbin holder 42. By rotating the turntable 40, the positions of the first bobbin holder 41 and the second bobbin holder 42 can be changed. It should be noted that as long as the positions of the first bobbin holder 41 and the second bobbin holder 42 can be changed, other devices can be used to replace the turntable 40. The first bobbin holder 41 can rotate relative to the turntable 40 with the axial position of the first bobbin holder 41 as the rotation center. 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 can rotate relative to the turntable 40 with the axial position of the second bobbin holder 42 as the rotation center. 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. Hereinafter, the first bobbin holder 41 and the second bobbin holder 42 are collectively referred to as bobbin holders 41, 42. The state where the bobbin holders 41, 42 are arranged one above the other is shown in FIG. 2. 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 thread winding machine 9 winds the thread 93 around the bobbin 91 of the bobbin holder 41, 42 at the winding position to manufacture the package 94. In addition, when the package 94 of the first bobbin holder 41 becomes full after winding a specified amount of the thread 93, the turntable 40 rotates, whereby the positions of the first bobbin holder 41 and the second bobbin holder 42 are switched. After that, the package 94 of the first bobbin holder 41 that has become full and is in the standby position is recovered, and the thread 93 is wound around the bobbin 91 of the second bobbin holder 42 at the winding position. A bobbin 91 is attached to the first bobbin holder 41 again after the package 94 has been recovered. The control device 50 includes a control unit 50a, a learning unit 50b, a determination unit 50c, a storage unit 50d, an input unit 50e, and an output unit 50f. Specifically described, the control device 50 is configured as a well-known computer, including a CPU (Central Processing Unit), a RAM (Random Access Memory), an SSD (Solid State Drive), etc. The CPU is a type of processor. Programs and data for controlling the wire winder 9 are pre-stored in the SSD. The CPU reads the program into the RAM and executes the program, thereby enabling the control device 50 to act as the control unit 50a, the learning unit 50b, and the determination unit 50c. In addition, the SSD corresponds to the storage unit 50d. It should be noted that an HDD (Hard Disk Drive) or a flash memory can also be used instead of the SSD. Alternatively, a memory provided outside the control device 50 and capable of communicating with the control device 50 can be used as the storage unit. Specifically, the input unit 50e and the output unit 50f are communication modules. The input unit 50e inputs data from the outside, and the output unit 50f outputs data to the outside. In this embodiment, the control device 50 corresponds to an abnormality determination device. That is, in this embodiment, the device for controlling the spinning and drawing machine 1 and the device for performing abnormality determination are shared. Instead, a device that performs abnormality determination and does not control the spinning and drawing machine 1 can also be set as the abnormality determination device. The control unit 50a is responsible for the overall control performed by the control device 50. The control unit 50a processes the data input from the outside to the input unit 50e of the control device 50 or the data stored in the storage unit 50d. The control unit 50a stores the data obtained through this processing in the storage unit 50d or outputs it to the outside of the control device 50. The learning unit 50b performs the process of constructing a determination model using machine learning. The details of the determination model constructed by the learning unit 50b will be described later. The determination unit 50c performs a determination process based on the data input from the outside to the control device 50 (specifically, the thermal image from the imaging device 6) and the determination model constructed by the learning unit 50b and stored in the storage unit 50d. The storage unit 50d stores data according to the processing of the control unit 50a. The notification unit 65 is shown in FIG. 3. 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. The abnormality that the control device 50 can determine is an abnormality related to the wire 93. Specifically, it is an abnormality in the wire path of the wire 93, an abnormality in the wire tension of the wire 93, or an abnormality in the adhesion amount of the sizing agent applied to the wire 93. In the present embodiment, as an example, an abnormality in the wire path of the wire 93 is determined. However, instead of or in addition to this, it is also possible to determine at least one of the above abnormalities. In FIG. 6(a), the wire path region and the state of the first wire guide 5a when the wire path is normal are shown. As shown in FIG. 6(a), the first wire guide 5a is a comb-shaped member. A plurality of guide grooves 61 for separately guiding a plurality of wires 93 are formed in the first wire guide 5a. When the wire path is normal, as shown in FIG. 6(a), one correct wire 93 enters one guide groove 61 and is guided. Here, when the wire 93 is guided by the first wire guide 5a, the wire 93 contacts the first wire guide 5a. Therefore, frictional heat is generated. The thermal image generated in this situation is shown in FIG. 6(b). As shown in FIG. 6(b), in this thermal image, the temperature of the contact portion between the wire 93 and the first wire guide 5a and its vicinity becomes higher than the surrounding temperature. Moreover, since the wire 93 enters all the guide grooves 61, the temperature of all the guide grooves 61 is higher than the surrounding temperature. On the other hand, in FIG. 7(a), the wire path region and the state of the first wire guide 5a when the wire path is abnormal are shown. FIG. 7(a) shows the following situation: Two wires 93 enter a certain guide groove 61 and are guided, and the wire 93 does not enter another guide groove 61. That is, a certain wire 93 is erroneously guided by an adjacent guide groove 61, so the wire path is abnormal. The thermal image generated in this situation is shown in FIG. 7(b). In FIG. 7(b), the temperature around the guide groove 61 into which two wires 93 enter is higher than the temperature of other guide grooves 61. Moreover, the temperature around the guide groove 61 that the wire 93 does not enter is lower than the temperature of other guide grooves 61. Thus, the abnormality of the wire path is clearly shown in the thermal image. Generally, since the wire 93 is thin, it is difficult to accurately determine the wire 93 in a visible image. However, since the frictional heat generated between the wire 93 and the guide groove 61 diffuses to the surroundings, the abnormality of the wire path (specifically, an abnormality related to whether the wire 93 enters the guide groove 61 or an abnormality related to whether the wire 93 deviates from the normal wire path during production) can be accurately detected by using the thermal image. In FIG. 4, the process for the learning unit 50b to construct a determination model using machine learning is conceptually shown. As shown in FIG. 4, the learning thermal image and its correct answer data are used as learning data for constructing the determination model. The learning thermal image refers to a thermal image generated by photographing the wire path region including the first wire guide 5a and used as learning data. In the present embodiment, the photographing device 6 is used to generate the learning thermal image. However, a device other than the photographing device 6 may also be used to generate the learning thermal image. The learning data is data obtained by establishing a correspondence between the learning thermal image and the correct answer data (label) indicating whether the wire path mapped in the learning thermal image is normal or abnormal. As the learning data, only the thermal images with normal wire paths can be used, only the thermal images with abnormal wire paths can be used, or both can be used. It is preferable that the learning thermal images included in the learning data are multiple. For example, the thermal images with normal wire paths are easy to capture (that is, it is easy to prepare the learning data). Although the thermal images with abnormal wire paths are time-consuming because the conditions of abnormal wire paths need to be intentionally created, by learning the thermal images with abnormal wire paths, it is possible to accurately determine whether the same type of abnormality occurs. The learning thermal image and the correct answer data are used for machine learning together. Therefore, the machine learning in this embodiment is supervised learning. The specific method of performing machine learning is arbitrary, but in this embodiment, deep learning is used. Thus, the learning unit 50b can extract features for determining the abnormality of the wire path from the thermal image by extracting or inferring the features on the thermal image in which the wire path is shown as normal or abnormal. For example, when learning multiple learning thermal images with normal wire paths, learn the features on the thermal images common to the multiple learning thermal images. Such features are, for example, the features that the temperatures of all the guide grooves 61 are uniform because the wire 93 enters each guide groove 61 of the first wire guide 5a and / or the highest temperature of the first wire guide 5a is high. In addition, when learning multiple learning thermal images with abnormal wire paths, learn the features on the thermal images common to the multiple learning thermal images. Such features are, for example, the features that the temperature is low because the wire 93 does not enter a certain guide groove 61 of the first wire guide 5a and / or the highest temperature of the first wire guide 5a is low. Through the above, the learning unit 50b constructs a determination model. The determination model constructed by the learning unit 50b is stored in the storage unit 50d. The above determination model is an example, and other methods can also be used to construct the determination model. That is, it is also possible to specify the feature quantities related to the abnormality of the wire path and perform machine learning instead of deep learning. In addition, either directly use the thermal images generated by photographing the wire path area for machine learning as described above, or perform machine learning after performing some preprocessing on the thermal images generated by photographing the wire path area. As the preprocessing, for example, image adjustment for clearly showing the wire 93 in the thermal image or deletion of objects other than the wire 93 and the first wire guide 5a can be considered. As shown in FIG. 5, at the stage of performing the abnormality determination of the wire path, the determination unit 50c performs the following processing using the determination model. That is, the determination model performs the abnormality determination of the wire path based on the thermal image generated by the imaging device 6. In the following description, the thermal image captured by the imaging device 6 at the stage of performing the abnormality determination of the wire path is referred to as the "thermal image for determination". As described above, the determination model is constructed by learning the features on the images of the wire path showing normal or abnormal states. Therefore, based on the thermal image for determination, the determination model determines whether there are such features, etc., and determines whether the wire path is normal or abnormal. It should be noted that even when an abnormality other than the wire path is set as the detection object, the same processing as when the abnormality of the wire path is set as the detection object can be performed. Hereinafter, the case where an abnormality other than the wire path can be determined based on the thermal image will be described. When the wire tension of the wire 93 is significantly high, the force of interaction between the wire 93 and the first wire guide 5a becomes large, so the frictional heat becomes large, and the temperature of the guide groove 61 tends to become high. Therefore, the abnormality of the wire tension of the wire 93 can be detected using the thermal image. For example, machine learning can be performed on the learning data obtained by establishing a correspondence between the thermal image captured in the normal state where the wire tension is within the normal range and the correct solution data of the normal wire tension. Or, machine learning can also be performed on the learning data obtained by establishing a correspondence between the thermal image captured in the abnormal state where the wire tension deviates from the normal range and the correct solution data of the abnormal wire tension. In addition, when the adhesion amount of the sizing agent applied to the wire 93 is too small, the friction coefficient between the wire 93 and the first wire guide 5a becomes high, so the frictional heat becomes large, and the temperature of the guide groove 61 tends to become high. Therefore, the abnormality of the adhesion amount of the sizing agent applied to the wire 93 can be detected using the thermal image. For example, machine learning can be performed on the learning data obtained by establishing a correspondence between the thermal image captured in the normal state where the adhesion amount of the sizing agent is within the normal range and the correct solution data of the normal adhesion amount of the sizing agent. Or, machine learning can also be performed on the learning data obtained by establishing a correspondence between the thermal image captured in the abnormal state where the adhesion amount of the sizing agent deviates from the normal range and the correct solution data of the abnormal adhesion amount of the sizing agent. In addition, when the wear of the first wire guide 5a intensifies, the friction coefficient of the first wire guide 5a becomes large, so the temperature of the guide groove 61 tends to become high. Therefore, the abnormality of the wear of the first wire guide 5a can be detected using the thermal image. For example, machine learning can be performed on the learning data obtained by establishing a correspondence between the thermal image captured in the normal state where the first wire guide 5a is not worn and the correct solution data indicating that the first wire guide 5a is not worn. Or, machine learning can also be performed on the learning data obtained by establishing a correspondence between the thermal image captured in the abnormal state where the first wire guide 5a is worn and the correct solution data indicating that the first wire guide 5a is worn. The determination model is created as a model capable of determining one of the above-mentioned abnormalities. However, the determination model can also be created as a model capable of determining multiple ones of the above-mentioned abnormalities. In this case, it may be that the determination model only determines whether at least one of the multiple abnormalities of the determination object has occurred (that is, it does not determine to what extent which abnormality has occurred). Or, it may also be that the determination model includes determining which one of the abnormalities of the determination object has occurred abnormally (in this case, it is necessary to learn the thermal images corresponding to the individual abnormal states). Next, mainly with reference to FIG. 8, the abnormality determination performed by the control device 50 will be described. The control device 50 determines whether it is the abnormality determination timing of the wire 93 (S101). The abnormality determination timing of the wire 93 is set, for example, at a specified time during the winding of the wire 93 at regular intervals. It should be noted that this is an example, and different settings can also be made as follows. For example, the abnormality determination timing of the wire 93 can also be set only at the start of the winding of the wire 93. Or, it can also be set to be the abnormality determination timing of the wire 93 when there is an operator's instruction. When the control unit 50a of the control device 50 determines that it is the abnormality determination timing of the wire 93, it instructs the imaging device 6 to capture an image of the wire path area (S102). Next, the determination unit 50c determines the abnormality related to the wire 93 as described above based on the determination model selected by the control device 50 and the determination thermal image captured by the imaging device 6 and input to the input unit 50e (S104). When the determination unit 50c does not determine that there is an abnormality related to the wire 93 (S105), the control device 50 performs the process of step S101 again. When the determination unit 50c determines that there is an abnormality related to the wire 93 (S105), the control unit 50a of the control device 50 controls the notification unit 65 to notify the operator of the abnormality (S106). By repeatedly performing the above process, the control device 50 determines whether the wire 93 is abnormal. In particular, by using the determination model constructed by machine learning for detection, the determination accuracy can be improved. In the present embodiment, the imaging device 6 captures the first wire guide 5a to generate a determination thermal image. Instead, the imaging device 6 can also capture the oil supply guide 3, the second wire guide 5b, the third wire guide 5c, or the traverse guide 23 to generate a determination thermal image. In addition, the wire guide of the determination object can be one or multiple. Hereinafter, for the sake of convenience of description, without particularly distinguishing the first wire guide 5a, the second wire guide 5b, and the third wire guide 5c, the first wire guide 5a, the second wire guide 5b, or the third wire guide 5c is referred to as the "wire guide 5". In the present embodiment, the determination unit 50c uses the thermal image for determining the wire path region to determine abnormalities. However, in addition to using the thermal image for determination, the determination unit 50c can also use the visible image for determining the wire path region to determine abnormalities. Specifically, a visible light camera is prepared and configured such that its lens faces the wire path region (specifically, the region including the first wire guide 5a). In particular, for the visible light camera, it is preferable that the viewing angle is close to the imaging device 6. The visible light camera captures the wire path region and generates a visible image for determination, and outputs it to the control unit 50a. As shown in FIG. 6(b), in the thermal image for determination, the temperature at the contact portion between the wire 93 and the guide groove 61 becomes the highest. In addition, when the wire path is normal, the wire 93 passes through a specified region of the guide groove 61. The specified region can be determined based on design information. Based on the above, by using the position of the first wire guide 5a (specifically, the position of the guide groove 61) and the information of the specified region obtained from the visible image for determination, the position through which the wire 93 passes when the wire path is normal can be determined in the visible image for determination. By applying this information to the thermal image for determination, the region that is likely to become high temperature when the wire path is normal can be determined. For example, by comparing the region that becomes the highest temperature in the thermal image for determination when the wire path is normal (the region where the wire 93 and the guide groove 61 contact on the thermal image for determination) with the actual temperature distribution of the thermal image for determination, it is possible to more accurately determine whether the wire path is normal. <B. Hardware Configuration of Control Device 50> Next, with reference to FIG. 9, the hardware configuration of the control device 50 shown in FIG. 3 will be described. FIG. 9 is a diagram showing an example of the hardware configuration of the control device 50. The control device 50 includes the above-mentioned storage unit 50d (refer to FIG. 3), a processor 81, a communication interface 84, a display interface 85, and an input interface 87. These components are connected to a bus BS. As the storage unit 50d, examples include a ROM (Read Only Memory), a RAM 83, and an auxiliary storage device 90, etc. The processor 81 is constituted by, for example, at least one integrated circuit. The integrated circuit can be constituted by, for example, at least one CPU, at least one GPU (Graphics Processing Unit), at least one ASIC (Application Specific Integrated Circuit), and at least one FPGA (Field Programmable Gate Array) or a combination thereof, etc. The processor 81 controls the operation of the control device 50 by executing various programs. The processor 81 reads the program to be executed from the auxiliary storage device 90 or the ROM 82 into the RAM 83 based on receiving execution commands for various programs. The RAM 83 functions as a working memory and temporarily stores various data required for program execution. A LAN (Local Area Network), an antenna, etc. are connected to the communication interface 84. The control device 50 exchanges data with external devices via the communication interface 84. Such external devices include, for example, servers. A display device 86 is connected to the display interface 85. The display interface 85 sends an image signal for displaying an image to the display device 86 according to instructions from the processor 81, etc. The display device 86 is, for example, a liquid crystal display, an organic EL (Electro Luminescence) display, or other displays. It should be noted that the display device 86 can be configured integrally with the control device 50 or separately from the control device 50. An input device 88 is connected to the input interface 87. The input device 88 is, for example, a mouse, a keyboard, a touch panel, or other devices that can accept user operations. It should be noted that the input device 88 can be configured integrally with the control device 50 or separately from the control device 50. The auxiliary storage device 90 is, for example, a storage medium such as a hard disk, a flash memory, and an SSD. The auxiliary storage device 90 stores, for example, the learning dataset 122, the determination model 124, the learning program 126, and the anomaly determination program 128. Their storage locations are not limited to the auxiliary storage device 90 and can also be stored in the storage area of the processor 81 (for example, a cache memory, etc.), the ROM 82, the RAM 83, external devices (for example, servers), etc. The learning program 126 is a program for generating the determination model 124 using the learning dataset 122. The learning program 126 can be provided in a manner of being embedded in a part of any program, rather than being provided as a single program. In this case, the learning process based on the learning program 126 is realized in cooperation with any program. Even a program that does not include such a part of the module is not a program that deviates from the gist of the learning program 126 of this embodiment. Moreover, one part or all of the functions provided by the learning program 126 can be realized by dedicated hardware. Also, the control device 50 can be configured in a form such as a so-called cloud server where at least one server executes a part of the processing of the learning program 126. The abnormality determination program 128 is a program for determining whether there is an abnormality in the wire passing through the wire guide 5 using the determined model 124 that has completed learning. The abnormality determination program 128 can be provided in a way that is embedded in a part of any program, rather than being provided as a standalone program. In this case, the determination process based on the abnormality determination program 128 is implemented in cooperation with any program. Even a program that does not include such a module part is not a program that deviates from the gist of the abnormality determination program 128 of this embodiment. Moreover, one part or all of the functions provided by the abnormality determination program 128 can be implemented by dedicated hardware. Moreover, the control device 50 can be configured in a form such as a so-called cloud server in which at least one server executes a part of the process of the abnormality determination program 128. <C. Learning dataset 122> Next, with reference to FIG. 10, the learning dataset 122 for generating the determined model 124 shown in FIG. 9 will be described. FIG. 10 is a diagram showing an example of the learning dataset 122. 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 from several tens to several hundreds of thousands. In each of the learning data 123, the category of the abnormality related to the wire passing through the above-mentioned wire guide 5 is associated as a label (correct value) with the thermal image indicating the distribution of the surface temperature of the wire guide. It should be noted that the learning data 123 can be obtained from one spinning and drawing machine 1 or from other spinning and drawing machines different from the spinning and drawing machine 1. The types of labels defined in the learning dataset 122 can be one type or multiple types. In the example of FIG. 10, abnormalities "A" to "D" are shown as labels indicating the categories of abnormalities. Abnormality "A" indicates a wire path abnormality, and this wire path abnormality indicates that the wire 93 deviates from the normal wire path area on the wire guide 5. Abnormality "B" indicates a tension abnormality, and this tension abnormality indicates that the tension of the wire 93 passing through the wire guide 5 deviates from the normal range. Abnormality "C" indicates an oil agent adhesion amount abnormality, and this oil agent adhesion amount abnormality indicates that the adhesion amount of the oil agent attached to the wire 93 passing through the wire guide 5 deviates from the normal range. Abnormality "D" indicates a wear abnormality, and this wear abnormality indicates that the wear amount of the wire guide 5 deviates from the normal range. It should be noted that a label indicating normal can also be associated with the thermal image. <D. Functional configuration of the control device 50> Next, with reference to FIGS. 11 to 13, the functional configuration of the control device 50 will be described. FIG. 11 is a diagram showing an example of the functional configuration of the control device 50. As shown in FIG. 11, as a functional configuration, the control device 50 includes a learning unit 50b, a determination unit 50c, an input unit 50e, and an output unit 50f. Hereinafter, these functional configurations will be described in the order of the learning unit 50b, the input unit 50e, the determination unit 50c, and the output unit 50f. (D1. Learning Unit 50b) First, referring to FIG. 12, the learning unit 50b shown in FIG. 11 will be described. FIG. 12 conceptually shows the learning process performed by the learning unit 50b. The learning unit 50b executes a learning process using the above-mentioned learning dataset 122 (refer to FIG. 10) to generate a determination model 124 for determining whether the wire 93 passing through the wire guide 5 is abnormal. The machine learning algorithm used in the learning process is not particularly limited. For example, deep learning, convolutional neural network (CNN), fully convolutional neural network (FCN), and support vector machine can be used. Hereinafter, the learning process using deep learning will be described. As shown in FIG. 12, the determination model 124 is composed of an input layer X, an intermediate layer H, and an output layer Y. The input layer X is configured to receive the input of the thermal image as the learning data 123. The input layer X is composed of, for example, N units x 1 to x N (N is a natural number). The number of units constituting the input layer X is the same as the number of dimensions of the input information. As an example, when the number of pixels of the thermal image is N pixels and each pixel of the thermal image is directly input to the input layer X, the input layer X is composed of N units. As another example, the feature amount extracted from the thermal image features can also be input to the input layer X. In this case, the input layer X is configured such that the number of its units is the same as the number of dimensions of the feature amount. Each unit constituting the input layer X outputs the input data to each unit of the first layer of the intermediate layer H. The intermediate layer H is composed of one or more layers. In the example of FIG. 12, the intermediate layer H is composed of L layers (L is a natural number). Each layer of the intermediate layer H includes a plurality of units. The number of units in each layer of the intermediate layer H can be the same or different. In the example of FIG. 12, the first layer of the intermediate layer H is composed of Q units h A1 to h AQ (Q is a natural number). In addition, the last layer of the intermediate layer H is composed of R units h L1 to h LR (R is a natural number). Each unit of each layer constituting the intermediate layer H is connected to each unit of the previous layer and each unit of the next layer. Each unit of each layer receives the output values from each unit of the previous layer, multiplies each output value by a weight, accumulates the results of their multiplications, adds (or subtracts) a specified deviation to the accumulated result, inputs the added result (or subtracted result) to a specified function (for example, a sigmoid function), and outputs the output value of the function to each unit of the next layer. The output layer Y outputs a推定 result corresponding to the input thermal image. The output layer Y is constituted by, for example, units y 1 to y 4 as follows. Hereinafter, the units y 1 to y 4 will also be referred to as unit y. Each of the units y is connected to the units h L1 to h LR of the last layer of the intermediate layer H. Each of the units y receives the output values from each unit of the last layer of the intermediate layer H, multiplies each output value by a weight, accumulates the results of their multiplications, adds (or subtracts) a specified deviation to the accumulated result, inputs the added result (or subtracted result) to a specified function (for example, a sigmoid function), and outputs the output result of the function as an output value. The number of units constituting the output layer Y is determined according to the number of abnormal categories specified in the learning data 123. As an example, in the case of detecting abnormalities "A" to "D", the number of units constituting the output layer Y is the units y 1 to y 4 These four. In this case, the unit y 1 is configured to output a score "sa" indicating the possibility of occurrence of abnormality "A". The unit y 2 is configured to output a score "sb" indicating the possibility of occurrence of abnormality "B". The unit y 3 is configured to output a score "sc" indicating the possibility of occurrence of abnormality "C". The unit y 4 is configured to output a score "sd" indicating the possibility of occurrence of abnormality "D". It should be noted that although in the example of FIG. 12, the configuration is such that one determination model 124 outputs multiple scores, it can also be configured such that one determination model outputs one score. As an example, it can be that the first determination model is configured to output a score "sa" related to anomaly "A", the second determination model is configured to output a score "sb" related to anomaly "B", the third determination model is configured to output a score "sc" related to anomaly "C", and the fourth determination model is configured to output a score "sd" related to anomaly "D". Next, an explanation will be given of the update process of the internal parameters of the determination model 124 performed by the learning unit 50b. The learning unit 50b inputs the thermal image specified in the first learning data 123 into the determination model 124. Next, the learning unit 50b compares the estimated results "sa" to "sd" output from the determination model 124 with the correct scores "sa'" to "sd'" corresponding to the anomaly categories established in the first learning data 123. As an example, when the anomaly category corresponding to in the learning data 123 is anomaly "A", the correct scores are (sa', sb', sc', sd') = (1, 0, 0, 0). When the anomaly category corresponding to in the learning data 123 is anomaly "B", the correct scores are (sa', sb', sc', sd') = (0, 1, 0, 0). When the anomaly category corresponding to in the learning data 123 is anomaly "C", the correct scores are (sa', sb', sc', sd') = (0, 0, 1, 0). When the anomaly category corresponding to in the learning data 123 is anomaly "D", the correct scores are (sa', sb', sc', sd') = (0, 0, 0, 1). The learning unit 50b calculates the error "Z" between the output results "sa" to "sd" of the determination model 124 and the correct scores "sa'" to "sd'". The error "Z" is calculated based on the following formula (1), for example. Z = {(sa - sa') 2 + (sb - sb') 2 + (sc - sc') 2 + (sd - sd') 2} / 4......(1) Next, the learning unit 50b updates various parameters (such as weights, biases) included in the determination model 124 to make the error "Z" smaller. The update of this parameter is realized by the error backpropagation method, for example. As described above, the learning unit 50b updates the internal parameters of the determination model 124 so that the estimation result output from the determination model 124 by inputting the thermal image included in the learning data 123 approaches the abnormal category that is the label associated with the thermal image. The learning unit 50b repeatedly performs the update process of the internal parameters of the determination model 124 for each learning data 123 included in the learning data set 122. As a result, the determination model 124 outputs a correct estimation result as learning progresses. It should be noted that the learning unit 50b does not need to use all the learning data 123 included in the learning data set 122 for the learning process, and can also generate the determination model 124 using a part of the learning data 123 included in the learning data set 122. The remaining learning data 123 is used for evaluation of the determination model 124, etc. (D2. Input unit 50e) Next, the function of the input unit 50e shown in FIG. 11 will be described. The input unit 50e is a communication module for realizing communication with the above-described imaging device 6. The control device 50 and the imaging device 6 can be communicatively connected either by wire or wirelessly. The input unit 50e obtains a determination thermal image from the imaging device 6 by outputting a shooting instruction to the imaging device 6, and inputs the determination thermal image to the determination model 124. At this time, the input unit 50e can either directly input each pixel of the determination thermal image to the determination model 124 or input the feature amount extracted from the determination thermal image to the determination model 124. Thus, the determination model 124 outputs a score indicating the possibility of an abnormality occurring in the wire 93 passing through the wire guide 5 according to the abnormal category. (D3. Determination unit 50c) Next, with reference to FIG. 13, the function of the determination unit 50c shown in FIG. 11 will be described. FIG. 13 conceptually shows the abnormality determination process performed by the determination unit 50c. As shown in FIG. 13, the determination model 124 outputs, for example, a score "sa" indicating the possibility of abnormality "A" occurring, a score "sb" indicating the possibility of abnormality "B" occurring, a score "sc" indicating the possibility of abnormality "C" occurring, and a score "sd" indicating the possibility of abnormality "D" occurring as the estimation result. The determination unit 50c determines the abnormal category that has occurred based on the scores "sa", "sb", "sc", and "sd". The determination unit 50c determines that abnormality "A" has occurred when the score "sa" exceeds the first threshold value. The first threshold value can be either preset or arbitrarily set by the user. When the score "sb" exceeds the second threshold value, the determination unit 50c determines that an abnormality "B" has occurred. The second threshold value can be set in advance or can be arbitrarily set by the user. In addition, the second threshold value can be the same as the above-mentioned first threshold value or can be different from the above-mentioned first threshold value. When the score "sc" exceeds the third threshold value, the determination unit 50c determines that an abnormality "C" has occurred. The third threshold value can be set in advance or can be arbitrarily set by the user. In addition, the third threshold value can be the same as the above-mentioned first and second threshold values or can be different from the above-mentioned first and second threshold values. When the score "sd" exceeds the fourth threshold value, the determination unit 50c determines that an abnormality "D" has occurred. The fourth threshold value can be set in advance or can be arbitrarily set by the user. In addition, the fourth threshold value can be the same as the above-mentioned first to third threshold values or can be different from the above-mentioned first to third threshold values. When the score "sa" is lower than the first threshold value, the score "sb" is lower than the second threshold value, the score "sc" is lower than the third threshold value, and the score "sd" is lower than the fourth threshold value, the determination unit 50c determines that no abnormality has occurred. That is, in this case, the determination unit 50c determines that the wire 93 passes through the wire guide 5 normally. (D4. Output unit 50f) Next, the function of the output unit 50f shown in FIG. 11 will be described. The output unit 50f outputs a control command corresponding to the determination result obtained by the determination unit 50c to a specified output destination. In one aspect, the output unit 50f outputs a display instruction for a warning to the above-mentioned display device 86 (refer to FIG. 9). As a result, the display device 86 displays a warning indicating that an abnormality has occurred in the wire guide 5. The warning displayed on the display device 86 includes, for example, the type of abnormality that has occurred and the thermal image for determination indicating the abnormality. It should be noted that the display instruction for the warning can also be sent to a pre-registered communication terminal. As a result, the person in charge and the manager can recognize that an abnormality has occurred in the wire guide 5 in the spinning and drawing machine 1. In other aspects, the output unit 50f outputs a notification instruction for a warning to the above-mentioned notification unit 65 (refer to FIG. 3). As a result, the notification unit 65 serving as a notification lamp emits light corresponding to the type of abnormality that has occurred. In addition, the notification unit 65 serving as a notification buzzer emits a sound corresponding to the type of abnormality that has occurred. As an example, when the determination result indicates normal, the output unit 50f causes the notification lamp to light up in green. In addition, when the determination result indicates an abnormality in the filament path, the output unit 50f causes the notification lamp to flash in red. Moreover, when the determination result indicates an abnormality in the amount of oil agent adhesion, the output unit 50f causes the notification lamp to flash in red. Moreover, when the determination result indicates an abnormality in the tension, the output unit 50f causes the notification lamp to flash in yellow. When the determination result indicates an abnormality in the wear of the godet 5, the output unit 50f causes the notification lamp to light up in yellow. In addition, it is also possible that the output unit 50f causes the notification buzzer to notify the abnormal state when the determination result indicates an abnormality in the filament path, an abnormality in the amount of oil agent adhesion, an abnormality in the tension, or an abnormality in the wear. Preferably, the output unit 50f causes the notification buzzer to output a sound corresponding to the type of abnormality. Moreover, it is also possible that the output unit 50f outputs a notification indicating the occurrence of an abnormality when the determination result indicates an abnormality in the filament path, an abnormality in the amount of oil agent adhesion, an abnormality in the tension, or an abnormality in the wear. This notification can be displayed on the above-mentioned display device 86 or sent to an external device. The content included in this notification, for example, includes the identifier of the spinning draw frame in which the abnormality occurred (for example, the machine ID) and the type of abnormality that occurred. Moreover, in other aspects, the output unit 50f outputs a storage instruction for the log to the above-mentioned storage unit 50d (refer to FIG. 3). Thereby, the storage unit 50d stores a log corresponding to the type of abnormality that occurred. As an example, this log includes the occurrence time of the abnormality, the type of abnormality that occurred, and a thermal image for abnormality determination indicating the abnormality. <E. First Modification Example> Next, with reference to FIG. 14, a first modification example of the above-described embodiment will be described. It should be noted that in the following description, for components that are the same as or similar to those in the foregoing embodiment, the same reference numerals may sometimes be marked in the drawings and the description may be omitted. In the example of FIG. 11 above, the learning unit 50b and the determination unit 50c are installed on the same spinning draw frame 1. However, the learning unit 50b and the determination unit 50c do not have to be installed on the same spinning draw frame 1. As an example, the learning unit 50b can also be installed on a different device. FIG. 14 is a diagram showing an example of the device configuration of the abnormality determination system 500 in this modification example. As shown in FIG. 14, the abnormality determination system 500 includes more than one spinning draw frame 1 and more than one information processing device 200. In the example of FIG. 14, the abnormality determination system 500 is composed of three spinning draw frames 1A to 1C and one information processing device 200. The spinning draw frames 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 spinning draw frame 1 and the information processing device 200 can be communicatively connected either by wire or by wireless. The information processing device 200 is a desktop personal computer, a notebook personal computer, a tablet terminal, or other information processing terminals. In the example of FIG. 14, the learning unit 50b is installed in the information processing device 200. In addition, the determination unit 50c is installed in the spinning and drawing machine 1C. The information processing device 200 collects the above-mentioned learning data 123 (refer to FIG. 10) from the spinning and drawing machines 1 (for example, spinning and drawing machines 1A and 1B) connected to the network NW. Next, the learning unit 50b of the information processing device 200 performs a learning process using the learning data 123 collected from all the spinning and drawing machines 1, and generates the above-mentioned determination model 124. Since the function of the learning unit 50b is as described above, its description will not be repeated. After that, the information processing device 200 sends the generated determination model 124 to the spinning and drawing machine 1 (for example, spinning and drawing machine 1C). The determination unit 50c uses the determination model 124 to determine whether an abnormality has occurred in the godet 5 in the spinning and drawing machine 1C. Since the function of the determination unit 50c is as described above, its description will not be repeated. <F. Second modification example> Next, with reference to FIGS. 15 and 16, a second modification example of the above-described embodiment will be described. The heat (temperature) generated by the friction between the wire 93 and the first godet 5a varies depending on the operating conditions of the spinning and drawing machine 1. For example, it shows the following tendency: as the traveling speed of the wire 93 increases, the temperature of the first godet 5a becomes higher. In addition, the friction coefficient varies depending on the material of the wire 93, and the temperature of the first godet 5a will change accordingly. Therefore, in this modification example, the above-mentioned determination model 124 is created according to each operating condition of the spinning and drawing machine 1. However, it is not necessary to create a determination model for each operating condition, and one determination model can also be used regardless of the operating conditions. (F1. Learning process) First, with reference to FIG. 15, the learning process of this modification example will be described. FIG. 15 is a diagram showing an example of the learning data set 122A of this modification example. The learning data set 122A includes a plurality of learning data 123A. The number of learning data 123A included in the learning data set 122A is arbitrary. As an example, the number of learning data 123A is from several tens to several hundreds of thousands. In the learning material 123 shown in FIG. 10 above, the abnormal category of the wire 93 is associated with a thermal image representing the distribution of the surface temperature of the wire guide 5 as a label (correct value). In contrast, in the learning material 123A of this modification example, the operating conditions of the spinning draw machine 1 or other spinning draw machines are further associated with the learning thermal image. As such operating conditions, for example, the traveling speed of the wire 93 and the material of the wire 93 can be cited. The above learning unit 50b classifies the learning materials 123A included in the learning material set 122A according to each similar operating condition. Any method can be adopted in the classification method. As an example, the learning material 123A is classified by clustering. The number of categories (classification number) can be preset or automatically determined. The method of clustering is not particularly limited, and any method can be used as long as it can classify similar operating conditions. For example, the method of clustering includes hierarchical methods and non-hierarchical methods. As a method of hierarchical clustering, the following method can be cited: on the space indicated by an index including at least one of the traveling speed of the wire 93 and the material of 93, two points with close distances are sequentially combined until the distance exceeds a predetermined value. In addition, as a method of non-hierarchical clustering, the k-means method etc. can be cited. The learning unit 50b executes the above learning process for each set of learning materials 123A belonging to the same category. The learning process is as described in FIG. 12 above, so its description will not be repeated. As a result of the learning process, determination models 124 are generated respectively according to the operating condition ranges of the spinning draw machine 1. Each of the determination models 124 is stored in a state associated with the operating condition range. (F2. Determination process) Next, with reference to FIG. 16, the determination process of this modification example will be described. FIG. 16 is a diagram showing a flowchart of the determination process of this modification example. In the flowchart shown in FIG. 16, compared with the flowchart shown in FIG. 8, the process of step S103 is added. Other aspects are as described in FIG. 8, so its description will not be repeated. In step S103, the above determination unit 50c acquires the current operating condition of the spinning draw machine 1. As such operating conditions, for example, the traveling speed of the wire 93 and the material of the wire 93 can be cited. Such operating conditions are input to the above input unit 50e, for example. Next, among the determination models 124 generated by the free learning unit 50b, the determination unit 50c acquires the determination model 124 corresponding to the current operating conditions of the spinning draw machine 1. More specifically, the determination unit 50c determines the operating condition range to which the current operating conditions belong from among the operating condition ranges associated with each of the determination models 124, and selects the determination model 124 corresponding to that operating condition range. After that, the determination unit 50c performs the above-described abnormality determination process using the selected determination model 124. The abnormality determination process performed by the determination unit 50c is as described in FIG. 13 above, and thus its description will not be repeated. (F3. Others) It should be noted that in the above content, an example of generating the determination model 124 according to each similar operating condition has been described, but the usage of the operating conditions of the spinning draw machine 1 is not limited to this. As an example, the above learning unit 50b performs the learning process including not only the thermal images included in the learning data 123A but also the operating conditions. In this case, the determination model 124 is configured to accept not only the input of the thermal images but also the input of the operating conditions. Next, the above input unit 50e inputs not only the determination thermal images but also the current operating conditions of the spinning draw machine 1 to the determination model 124. Thereby, the determination model 124 more accurately outputs a score indicating the possibility of an abnormality occurring in the wire 93 passing through the wire guide 5. <G. Third modification example> Next, with reference to FIGS. 17 and 18, a third modification example of the above-described embodiment will be described. In FIGS. 15 and 16 above, an example of generating the determination model 124 according to each similar operating condition has been described, but the determination model 124 can also be generated according to each other information different from the operating conditions. As an example, the determination model 124 can be generated according to the shooting conditions of the shooting device 6 for the wire guide 5, respectively. Such shooting conditions include, for example, the relative positional relationship of the shooting device 6 with respect to the wire guide 5. Hereinafter, an example in which the determination model 124 is generated according to the relative positional relationship of the shooting device 6 with respect to the wire guide 5, respectively, will be described. (G1. Learning process) First, with reference to FIG. 17, the learning process of this modification example will be described. FIG. 17 is a diagram showing an example of the learning data set 122B of this modification example. The learning data set 122B includes a plurality of learning data 123B. The number of the learning data 123B included in the learning data set 122B is arbitrary. As an example, the number of the learning data 123B is from several tens to several hundreds of thousands. In the learning material 123 shown in FIG. 10 above, the abnormal category of the wire 93 is associated with the thermal image representing the surface temperature distribution of the wire guide 5 as a label (correct value). In contrast, in the learning material 123B of this modification, information indicating the positional relationship between the wire guide 5 and the imaging device 6 is further associated with the learning thermal image. This positional relationship is defined by, for example, at least one of the relative coordinate values of the imaging device 6 with respect to the wire guide 5 and the viewing angle of the imaging device 6 with respect to the wire guide 5. The positional relationship included in the learning material 123 may be the positional relationship between the wire guide 5 and the imaging device 6 provided in the spinning and drawing machine 1, or may be the positional relationship between the wire guide 5 and the imaging device 6 provided in another model different from the spinning and drawing machine 1. The learning unit 50b classifies the learning materials 123B included in the learning material set 122B according to each similar positional relationship. Any method can be adopted in the classification method. As an example, the learning material 123B is classified by clustering. The number of categories (classification number) can be either preset or automatically determined. The clustering method is not particularly limited, and any method can be used as long as it can classify similar positional relationships. For example, the clustering method includes hierarchical methods and non-hierarchical methods. As a hierarchical clustering method, the following method can be cited: On the space indicated by an index including at least one of the relative coordinate values of the imaging device 6 with respect to the wire guide 5 and the viewing angle of the imaging device 6 with respect to the wire guide 5, two points with close distances are sequentially combined until the distance exceeds a predetermined value. In addition, as a non-hierarchical clustering method, the k-means method etc. can be cited. The learning unit 50b executes the above learning process for each set of learning materials 123B belonging to the same category. The learning process is as described in FIG. 12 above, so its description will not be repeated. As a result of the learning process, determination models 124 are generated according to similar positional relationships respectively. Each of the determination models 124 is stored in a state where the positional relationship is associated. (G2. Determination process) Next, referring to FIG. 18, the determination process of this modification will be described. FIG. 18 is a diagram showing a flowchart of the determination process of this modification. In the flowchart shown in FIG. 18, compared with the flowchart shown in FIG. 8, the process of step S103A is added. Other aspects are as described in FIG. 8, so their description will not be repeated. In step S103A, the determination unit 50c acquires the positional relationship between the wire guide 5 and the imaging device 6. This positional relationship can be either preset or input by the user. This positional relationship is input to the input unit 50e above, for example. Hereinafter, the positional relationship between the wire guide 5 and the imaging device 6 will also be referred to as the "input positional relationship". Next, among the determination models 124 generated by the free learning unit 50b of the determination unit 50c, the determination model 124 corresponding to the above input positional relationship is acquired. More specifically, the determination unit 50c determines the positional relationship to which the above input positional relationship belongs from among the positional relationships associated with each of the determination models 124, and selects the determination model 124 corresponding to that positional relationship. After that, the determination unit 50c performs the above abnormality determination process using the selected determination model 124. The abnormality determination process performed by the determination unit 50c is as described in FIG. 13 above, and thus its description will not be repeated. (G3. Others) It should be noted that in the above content, an example of generating the determination model 124 according to each similar positional relationship has been described, but the usage of this positional relationship is not limited thereto. As an example, the above learning unit 50b performs the learning process including not only the thermal images included in the learning data 123B but also the above positional relationship. In this case, the determination model 124 is configured to accept not only the input of the thermal image but also the input of the above positional relationship. Next, the above input unit 50e inputs not only the determination thermal image to the determination model 124 but also the positional relationship between the wire guide 5 and the imaging device 6 to the determination model 124. Thereby, the determination model 124 more accurately outputs a score indicating the possibility that the wire 93 passing through the wire guide 5 is abnormal. <H. Fourth Modification>[ Next, with reference to FIG. 19, a fourth modification of the above embodiment will be described. FIG. 19 is a diagram schematically showing the shearing process of this modification. When the relative positional relationship between the wire guide 5 and the imaging device 6 is fixed, the position of the wire guide 5 mapped in the thermal image is also fixed. In this case, the control device 50 can improve the learning accuracy and the determination accuracy by shearing a predetermined area in the thermal image and using the image information of the sheared area for the learning process and the determination process. However, the position of the imaging device 6 relative to the wire guide 5 may also vary depending on the model of the spinning draw machine 1. Therefore, the control device 50 of this modification first determines the area of the wire guide 5 in the thermal image. More specifically, inside the spinning draw machine 1, in addition to the imaging device 6 which is a thermal imaging camera, a visible light camera (not shown) is also provided. The visible light camera is provided inside the spinning draw machine 1 such that its shooting field of view includes the wire guide 5. Preferably, the visible light camera is provided adjacent to the imaging device 6. In addition, the angle formed by the optical axis of the visible light camera and the optical axis of the imaging device 6 is lower than a specified angle (for example, lower than 10°). As an example, the specified angle is 0°. In this case, the optical axis of the visible light camera is parallel to the optical axis of the imaging device 6. The visual image IM1 obtained from the visible light camera and the thermal image IM2 obtained from the self-shooting device 6 are shown in Figure 19. The control device 50 determines the image region R1 of the mapping thread guide 5 from the visual image IM1. Then, the image area R2 mapped by the thread guide 5 is determined within the thermal image IM2 based on the position of the image area R1 within the visual image IM 1 The control device 50 sets the position of the image area R2 within the thermal image IM2 to be the same as the position of the image area R1 within the visual image IM1. In other states, the control device 50 performs a projection transformation on the image area R1 within the visual image IM1 based on a predetermined projection matrix, which determines that the region after the transformation is determined as the image area R2 within the thermal image IM2. Thereafter, the image area R2 within the control device 50 shear thermal image IM2, producing a shear image IM3 can be used for the learning processing performed by the aforementioned learning section 50b In addition, the shear image IM3 can be used for the judgment processing performed by the above judgment section 50c Then, with reference to FIG. The dummy jig processing machine 100 processes the joint fiber yarn 104 as shown in FIG. The thread supply section 101 has a mechanism to keep the plurality of rolls wound with the fitted fiber threads 104 . The machining section 102 performs false-prong processing on the joint fiber yarn 104 Device 107 , cooling device 108 , dummy clamp device 109 , second feed roller 110 , merging device 111 , third feed roller 112 , second processing device 113 , and fourth feed roller 114 . The first feed roller 105 is composed of opposing rollers that pull out the synthed fiber strands 104 by clamping and rotating a plurality of coiled fibers 104 , respectively. The thread guide 106 , respectively, guides a plurality of joint fiber threads 104 pulled out of the first feed roller 105 . The first heating device 107 is, for example, equipped with a heater to heat a plurality of synthed fibers 104 . In addition, a thread guide omitting the icon is provided within the first heating device 107 . The cooling device 108 , for example, uses an airflow to cool the plurality of synthed fibers 104 . Furthermore, a thread guide omitting the icon is provided within the cooling device 108 . The dummy device 109 performs dummy on multiple synfiber filaments 104 . A second feed roller 110 pulls out a plurality of synthed fiber strands 104 that have been implemented by the dummy device 109 . The merging device 111 performs merging of two synthed fiber yarns 104 . A third feed roller 112 pulls out a plurality of combined fiber strands 104 that have been merged by the merging device 111 . The second processing device 113 heats the plurality of synthed fibers 104 again. The fourth feed roller 114 pulls out the plurality of synthed fiber strands 104 that have been overheated by the second processing device 113 . The winding portion 103 is made by winding a plurality of composite fiber threads 104 pulled from the fourth feed roller 114 , respectively, to fabricate a plurality of windings. Configured near the thread guide 106 has a shooting device 6 equivalent to the above embodiment. The shooting device 6 is configured in such a way that the lens is directed toward the wire guide 106 to capture the wire guide 106 and generate a thermal image for judgment. High friction heat is easily generated due to the application of a higher thread tension to the bonded fiber strand 104 in the vicinity of the thread guide 106 . Thus, the same processing as in the above embodiments by resorting to thermal imagery based on the thread guide 106 enables the detection of abnormalities in the thread channel. The shooting device 6 may also capture a portion other than the thread guide 106 to generate a thermal image for judgment. For example, a thread guide inside the first heating device 107 , a thread guide inside the cooling device 108 , or a traverse guide provided in the winding portion 103 , or the like, may also be photographed to generate the thermal image for judgment. <J. Summary> As explained above, the control device 50 of this embodiment has an input portion 50 e , a judgment portion 50 c , and an output portion 50 f inputting a thermal image (input process, input unit) including a thread channel region that is a first thread guide 5a provided by a spinning traction machine 1 (or dummy processor 100 , the same below) part of (thread guide 106 , the same below) In addition, the control device 50 performs an anomaly determination method including an input process, a judgment process, and an output process. In addition, an anomaly determination program useful for making the control device 50 function as an input unit, a judgment unit, and an output unit is stored in the storage portion 50d of the control device 50 . The friction conditions generated at the first thread guide 5a are different when the thread 93 is normal and when the thread 93 is abnormal, and therefore the temperature of the first thread guide 5a is different. Furthermore, even in cases where the thread 93 is too small and makes the detection of the thread 93 itself difficult, the thread 93 can be detected indirectly as long as the thermal image is due to frictional thermal anomaly. In the control device 50 of this embodiment, the judgment portion 50c determines an abnormality of the thread channel based on a thermal image from the input to the input portion 50e. In the case of thread channel anomalies, the temperature of the first thread guide 5a changes, thus determining the presence or absence of thread channel anomalies with high precision. In the control device 50 of this embodiment, the judgment portion 50c determines an abnormality in the wire tension based on a thermal image from the input to the input portion 50e. In the case of abnormal thread tension, the temperature of the first thread guide 5a changes, thus determining the presence or absence of abnormal thread tension with high precision. In the control device 50 of this embodiment, the judgment portion 50c determines an abnormality in the amount of adhesion imparted to the thread 93 based on a thermal image from the input to the input portion 50e. In the case of an abnormal amount of attachment of the oil agent imparted to the thread 93 , the temperature of the first thread guide 5a changes, thus determining with high precision whether there is an abnormality in the amount of the oil imparted to the thread 93 . The control device 50 of this embodiment stores a plurality of judgment models corresponding to the operating conditions The judgment section 50 c determines whether there is an abnormality related to the thread 93 based on the judgment model selected according to the current operating condition and the thermal image input to the input section 50 e . The temperature that can be generated in the first thread guide 5a varies according to the operating conditions, so anomalies related to the thread 93 can be judged with high precision by using a judgment model corresponding to the operating conditions. In the control device 50 of this embodiment, a thermal image of the input to the input portion 50e is taken by the capture device 6 . Thereby, thermal images including the silk tract region can be obtained and used for the judgment of anomalies related to the silk 93 . The output of the control device 50 of this embodiment serves as a judgment result of an anomaly related to the thread 93 of the spinning traction machine 1 of the synthetic fiber manufacturing equipment. The output of the control device 50 of the other embodiments described above serves as a judgment result of anomalies in the synth yarn 104 of the dummy jib processing machine 100 of the synthetic fiber manufacturing equipment. From this, an anomaly related to yarn 93 (or synthed yarn 104) of the spinning traction machine 1 (or pseudo-pin processing machine 100) can be determined. In the control device 50 of this embodiment, the judgment model is constructed by using the learning-use thermal images generated by pre-photographing the thread channel region as learning-use data for machine learning. The judgment model is the model used to determine whether the thread 93 is abnormal. A better embodiment of the invention has been described above, but the above formulation may, for example, be changed as follows. In the above embodiment, the judgment unit 50c determines an anomaly related to the thread 93 using the judgment-use thermal image and the judgment model. Two models can also be used for that treatment. Also, in creating the judgment model, the learning section 50b constructs the first judgment model by using the thermal image of learning and the positive solution data representing the temperature judgment site (specifically, at and near the site where the thread 93 contacts the first thread guide 5a in normal times) as learning data. Then, the learning part 50b uses the temperature distribution of the temperature judgment site and the positive solution data indicating whether the thread 93 or the first thread guide 5a is abnormal as learning data to construct the second judgment model. Then, at the time of judgment, the judgment part 50c determines the range of the temperature judgment site based on the thermal image of the judgment and the first judgment model. Then, the judgment part 50c determines whether the thread 93 or the first thread guide 5a is abnormal based on the range of the temperature judgment site and the second judgment model, and outputs the judgment result. The flowchart shown in the above embodiments is an instance where a portion of the processing may be omitted or the content of a portion of the processing may be changed or a new processing may be appended. For example, in the flow diagram of FIG. <I. Supplementary Explanation> As described above, this embodiment comprises revealing content as follows. The problem to be solved by the present invention is as described above. Next, the method for solving the problem and its effects will be described. According to a first aspect of the present invention, there is provided an abnormality determination device having the following configuration. That is, the abnormality determination device includes an input unit, a determination unit, and an output unit. A thermal image including a thread path region is input to the input unit, and the thread path region is a part of a thread guide provided in a synthetic fiber manufacturing apparatus. The determination unit determines whether there is an abnormality related to the thread based on the thermal image input to the input unit. The output unit outputs the determination result of the determination unit. When the thread is normal and abnormal, the friction conditions generated in the thread guide are different, so the temperature of the thread guide is different. In addition, even when it is difficult to detect the thread itself because the thread is too thin, the thread can be indirectly detected as long as it is a thermal image due to the diffusion of frictional heat. By the above, it is possible to accurately determine an abnormality related to the thread based on a thermal image including a thread path region that is a part of a thread guide. In the above abnormality determination device, preferably, the determination unit determines whether there is an abnormality in the thread path based on the thermal image input to the input unit. In the case of an abnormality in the thread path, the temperature of the thread guide changes, so it is possible to accurately determine whether there is an abnormality in the thread path. In the above abnormality determination device, preferably, the determination unit determines whether there is an abnormality in the thread tension based on the thermal image input to the input unit. In the case of an abnormal thread tension, the temperature of the thread guide changes, so it is possible to accurately determine whether there is an abnormality in the thread tension. In the above abnormality determination device, preferably, the determination unit determines whether there is an abnormality in the adhesion amount of the sizing agent applied to the thread based on the thermal image input to the input unit. In the case of an abnormal adhesion amount of the sizing agent applied to the thread, the temperature of the thread guide changes, so it is possible to accurately determine the abnormality in the adhesion amount of the sizing agent. In the above abnormality determination device, preferably, it is configured as follows. That is, the abnormality determination device includes a plurality of the above determination models corresponding to the operating conditions. The determination unit determines whether there is an abnormality related to the thread based on the determination model selected according to the current operating conditions and the thermal image input to the input unit. The temperature generated in the thread guide can change according to the operating conditions. Therefore, by using a determination model corresponding to the operating conditions, it is possible to accurately determine an abnormality related to the thread. In the above abnormality determination device, preferably, the thermal image input to the input unit is captured by a capturing device. Thereby, a thermal image including a thread path region can be obtained and used for determining an abnormality related to the thread. In the above-mentioned anomaly judgment apparatus, it is preferable to output as the judgment result of the anomalies related to the threads of the spinning traction machine of the above-mentioned synthetic fiber manufacturing equipment. In the above-mentioned anomaly judgment apparatus, it is preferable to output as the judgment result of the anomalies related to the threads of the dummy-prong processing machine of the above-mentioned synthetic fiber manufacturing equipment. From this, the abnormalities related to the threads of the spinning traction machine or the dummy cock processing machine can be determined. According to a second view of the invention, a method for the following anomaly determination is provided. That is, the anomaly determination method includes the input process, the judgment process, and the output process. In the above input process, the input includes a thermal image of the thread channel region, which is part of the thread guide equipped with the synthetic fiber manufacturing equipment. In the above judgment process, based on the above thermal image input in the above input process, whether there are any abnormalities related to the threads is determined. In the above output process, the judgment result in the above judgment process is output. According to a third view of the invention, an anomaly determination program for making a computer function as the following unit is provided. The units include an input unit, a judgment unit, and an output unit. In the above-mentioned input unit, the input includes a thermal image of the thread channel region, which is part of the thread guide equipped with the synthetic fiber manufacturing equipment. In the above judgment unit, based on the above thermal image input in the above input unit, whether there are any abnormalities related to the threads is determined. In the above output unit, output the judgment result in the above judgment unit. It should be held that the embodiment of this elucidation is exemplary rather than restrictive in all respects. The categories of the present invention are shown by the category for which the scope of the patent is applied, and not by the foregoing explanation, intended to include meanings equivalent to the category for which the scope of the patent is applied and all variations within the category. 1: Spining Traction Machine (Synthetic Fiber Manufacturing Equipment) 1A: Spinning Traction Machine 1B: Spinning Traction Machine 1C: Spinning Traction Machine 2: Spinning Device 2a: Spinning Head 3: Oil Supply Guide 4: Wire Cutting Suction Device 4a: Driver Device 4b: Source of Attraction 5: Wire Guide 5a: Third Wire Guide (5b: 2nd Wire Guide) 6: Shooting Device 7: First guide roller 8: Second guide roller 9: Wire winder 11: Frame 20: First shell 21: Transverse device 22: Transverse cam 23: Transverse guide 30: Second housing 31: Contact roller 32: Operating panel 40: 1st Bracket tube 414: Turntable tube holder 50a: Control 50b: Learning 50c: Judgment 50d: Storage 50e: Input 50f: Output 51: Transverse motor 53: Turntable motor 54: First tube bracket motor 84: Communication Interface 85: Display Interface 86: Display Equipment 87: Input Interface 88: Input Equipment 90: Auxiliary Storage Device 91: Cylinder Tube 93: Wire 94: Rolling 100: Dummy Machine Processor (Synthetic Fiber Manufacturing Equipment) 106: Wire guide 107: First heating device 108: Cooling device 109: False cock device 110: Second feed roller 111: Wire merging device 112: Third feed roller 113: Second processing device 114: Fourth feed roller 122: Learning with dataset 122A Learning 12 with dataset 12: Learning with data set 122A Learning with dataset 12 123A: Data for Learning 123B: Data for Learning 124: Judgment Model 126: Learning Program 128: Anomaly Judgment Program 200: Information Processing Device 500: Anomaly Judgment System BS: Bus IM1: Visual Image IM2: Thermal Image IM3: Cut image NW: Network R1: Image Area R1: Image Step Area S103 Step: Image Step Area S103 S103A: Step S104: Step S105: Step S106: Step is a diagram representing a side view of a spinning traction machine of one embodiment of the present invention. 2 is a diagram representing the main view of the thread winder. 3 is a diagram representing a block diagram of a thread winder. Figure 4 represents the processing of the construction of the judgment model by means of machine learning. Fig. 5 is a plot representing the processing of the judgment result obtained using the judgment model. Figure 6(a), (b) shows the condition and thermal images of the thread channel area and the thread guide when the thread and the thread guide were normal. Figure 7(a), (b) shows the condition and thermal images of the thread channel region and the thread guide when the thread or thread guide is abnormal. Figure 8 is a diagram representing the flowchart of the anomaly judgment of the threads using the judgment model. 9 is a diagram representing one instance of the hardware configuration of the control device. Figure 10 is a graph representing an instance of the learning dataset. 11 is a diagram representing one instance of the functional configuration of the control device. Figure 12 is a diagram that conceptually represents the learning processing carried out by the Learning Department. Figure 13 is a diagram that conceptually represents the anomaly judgment processing performed by the judgment department. 14 is a diagram representing one instance of the device configuration of an anomaly judgment system. Figure 15 is a plot representing an instance of the learning dataset. Figure 16 is a diagram representing the flowchart of the judgment processing. Figure 17 is a graph representing an instance of the learning dataset. Figure 18 is a diagram representing the flowchart of the judgment processing. Figure 19 presents a rough representation of the shear treatment. 20 is a graph representing a side view of the dummy cock processing machine. 1: Spinning traction machine 2: spinning device 2a: Spray head 3: Oil supply guide 4: Silk cut off the suction device 4a: Drive device 4b: Source of attraction 5a: First thread guide 5b: Second thread guide 5c: Third thread guide 6: Shooting device 7: First guide thread roller 8: Second guide thread roller 9: Thread winder 11: Framework 20: First outer shell 40: Switch board 41: First tube bracket 42: Second tube bracket 91: tube 93: Thread 94: Rolls

Claims

1. An anomaly determination device, characterized by comprising: an input unit for inputting a thermal image including a yarn path region, wherein the yarn path region is part of a yarn guide provided by a synthetic fiber manufacturing equipment; a determination unit for determining whether there is an anomaly related to the yarn based on the thermal image input to the input unit; and an output unit for outputting the determination result of the determination unit.

2. As in request item 1, the anomaly determination device, wherein, The aforementioned determination unit determines whether or not the aforementioned anomaly exists based on the aforementioned thermal image used for determination input to the aforementioned input unit and the determination model generated by machine learning on multiple learning data. Among the aforementioned multiple learning data, the presence or absence of anomalies related to the yarn passing through the yarn guide provided by the aforementioned synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment is used as a label and associated with the learning thermal image mapped to the yarn guide.

3. As in request item 2, the anomaly determination device, wherein, Among the various learning materials mentioned above, the categories of the aforementioned anomalies were used as the aforementioned labels and associated with the aforementioned thermal images used for learning.

4. The anomaly detection device as described in request item 2, wherein, The aforementioned anomaly determination device further includes a learning unit for generating the aforementioned determination model based on the aforementioned multiple learning data.

5. The anomaly detection device as described in request item 3, wherein, The aforementioned anomaly determination device further includes a learning unit for generating the aforementioned determination model based on the aforementioned multiple learning data.

6. The anomaly detection device as described in request item 1, wherein, The above-mentioned anomalies include those related to the silk path.

7. The anomaly detection device as described in request item 2, wherein, The above-mentioned anomalies include those related to the silk path.

8. The anomaly detection device as described in request item 3, wherein, The above-mentioned anomalies include those related to the silk path.

9. The anomaly detection device as described in request item 4, wherein, The above-mentioned anomalies include those related to the silk path.

10. The anomaly detection device as described in request item 5, wherein, The above-mentioned anomalies include those related to the silk path.

11. The anomaly determination device as described in request item 1, wherein, The above-mentioned abnormalities include abnormalities in the tension of the yarn.

12. The anomaly determination device as described in request item 2, wherein, The above-mentioned abnormalities include abnormalities in the tension of the yarn.

13. The anomaly determination device as described in request item 3, wherein, The above-mentioned abnormalities include abnormalities in the tension of the yarn.

14. The anomaly detection device as described in claim 4, wherein, The above-mentioned abnormalities include abnormalities in the tension of the yarn.

15. The anomaly detection device as described in request item 5, wherein, The above-mentioned abnormalities include abnormalities in the tension of the yarn.

16. The anomaly determination device as described in request item 1, wherein, The above-mentioned abnormalities include abnormalities in the amount of oil applied to the silk thread.

17. The anomaly determination device as described in claim 2, wherein, The above-mentioned abnormalities include abnormalities in the amount of oil applied to the silk thread.

18. The anomaly determination device as described in request item 3, wherein, The above-mentioned abnormalities include abnormalities in the amount of oil applied to the silk thread.

19. The anomaly detection device as described in claim 4, wherein, The above-mentioned abnormalities include abnormalities in the amount of oil applied to the silk thread.

20. The anomaly detection device as described in claim 5, wherein, The above-mentioned abnormalities include abnormalities in the amount of oil applied to the silk thread.

21. The anomaly determination device as described in claim 2, wherein, The aforementioned anomaly determination device has multiple determination models learned from the operating conditions of similar synthetic fiber manufacturing equipment. The determination unit selects a determination model from the multiple determination models that corresponds to the current operating conditions of the synthetic fiber manufacturing equipment. Based on the selected determination model and the thermal image for determination input to the input unit, it determines whether the aforementioned anomaly exists.

22. The anomaly detection device as described in request item 3, wherein, The aforementioned anomaly determination device has multiple determination models learned from the operating conditions of similar synthetic fiber manufacturing equipment. The determination unit selects a determination model from the multiple determination models that corresponds to the current operating conditions of the synthetic fiber manufacturing equipment. Based on the selected determination model and the thermal image for determination input to the input unit, it determines whether the aforementioned anomaly exists.

23. The anomaly detection device as described in claim 4, wherein, The above-mentioned anomaly judgment device has a number of the above-mentioned judgment models obtained by learning separately according to the similar operating conditions of the above-mentioned synthetic fiber manufacturing equipment.

24. The anomaly detection device as described in request item 5, wherein, The above-mentioned anomaly judgment device has a number of the above-mentioned judgment models obtained by learning separately according to the similar operating conditions of the above-mentioned synthetic fiber manufacturing equipment.

25. An anomaly detection device as described in any of requests 1 to 20, wherein, The above-mentioned synthetic fiber manufacturing equipment is then equipped with a shooting device configured to capture the said thread guide, from which the above-mentioned shooting device obtains the said thermal image input to the said input portion.

26. The anomaly detection device as described in claim 25, wherein, The above anomaly judgment device has a number of the above judgment models obtained by learning the positional relationship between the thread guide and the shooting device equipped with the above synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment.

27. An anomaly detection device as described in any of requests 1 to 24, wherein, The above mentioned synthetic fiber manufacturing equipment is a spinning traction machine.

28. The anomaly detection device as described in claim 25, wherein, The above mentioned synthetic fiber manufacturing equipment is a spinning traction machine.

29. An anomaly detection device as described in any of requests 1 to 24, wherein, The aforementioned synthetic fiber manufacturing equipment is a dummy processing machine.

30. The anomaly detection device as described in claim 25, wherein, The aforementioned synthetic fiber manufacturing equipment is a dummy processing machine.

31. An anomaly detection method, characterized by comprising: steps to input a thermal image of the thread channel region, which is part of a thread guide equipped with synthetic fiber manufacturing equipment, steps to determine whether there are abnormalities related to the thread based on the above thermal images input in the above input steps; 32. As in request item 31, the exception determination method, wherein, In the steps of the above judgment, the presence or absence of the above anomalies is determined based on the above thermal image for the judgment entered in the above input step and the judgment model generated by means of machine learning on multiple learning data, each of the above multiple learning data, the presence or absence of the anomaly related to the thread that passes through the thread guides that pass through the said synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment is used as a label and maps the thread guide for learning.

33. As in request item 32, the exception handling method, wherein, Among each of the above plurality of learning-use data, associations were made with the above-mentioned categories of anomalies as the above-mentioned labels with the above-mentioned thermal images for learning use.

34. As in request item 32, the exception handling method, wherein, The above anomaly judgment method then has the step of generating the above judgment model based on the above multiple learning data.

35. The exception handling method as described in request item 33, wherein, The above anomaly judgment method then has the step of generating the above judgment model based on the above multiple learning data.

36. The exception determination method as described in any of requests 31 to 35, wherein, The above anomalies include those of the silk tract.

37. The exception determination method as described in any of requests 32 to 35, wherein, The above-mentioned anomaly determination method further includes the step of acquiring multiple determination models learned according to the operating conditions of similar synthetic fiber manufacturing equipment. The determination step includes: selecting a determination model from the multiple determination models that corresponds to the current operating conditions of the synthetic fiber manufacturing equipment; and determining whether there is an anomaly based on the selected determination model and the thermal image used for determination input in the above-mentioned input step.

38. The exception determination method as described in any of requests 31 to 35, wherein, The aforementioned synthetic fiber manufacturing equipment further includes an imaging device configured to capture images of the aforementioned guide fiber, and to acquire the aforementioned thermal image input during the aforementioned input step from the imaging device.

39. The exception handling method as described in request item 36, wherein, The aforementioned synthetic fiber manufacturing equipment further includes an imaging device configured to capture images of the aforementioned guide fiber, and to acquire the aforementioned thermal image input during the aforementioned input step from the imaging device.

40. As in request item 38, the exception handling method, wherein, The aforementioned anomaly determination method further includes the step of acquiring multiple determination models obtained by learning the positional relationship between the guide wire and the imaging device of the synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment. The determination step includes: selecting a determination model from the multiple determination models that corresponds to the positional relationship between the guide wire and the imaging device of the synthetic fiber manufacturing equipment; and determining whether there is an anomaly based on the selected determination model and the thermal image for determination input in the input step.

41. As in request item 39, the exception handling method, wherein, The aforementioned anomaly determination method further includes the step of acquiring multiple determination models obtained by learning the positional relationship between the guide wire and the imaging device of the synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment. The determination step includes: selecting a determination model from the multiple determination models that corresponds to the positional relationship between the guide wire and the imaging device of the synthetic fiber manufacturing equipment; and determining whether there is an anomaly based on the selected determination model and the thermal image for determination input in the input step.

42. An exception detection program, wherein, The above-mentioned anomaly determination program causes the anomaly determination device to perform the following steps: inputting a thermal image of a yarn path region, wherein the yarn path region is part of a yarn guide provided by a synthetic fiber manufacturing equipment; determining whether there is an anomaly related to the yarn based on the thermal image input in the above-mentioned input step; and outputting the determination result of the above-mentioned determination step.

43. As in the exception handling procedure of request item 42, wherein, In the above-mentioned determination step, based on the thermal image used for determination input in the above-mentioned input step and the determination model generated by machine learning on multiple learning data, it is determined whether the above-mentioned anomaly exists. Among the multiple learning data, the presence or absence of anomalies related to the yarn passing through the guide of the above-mentioned synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment is used as a label and associated with the learning thermal image mapping the guide.

44. As in request item 43, the exception handling procedure, wherein, Among the various learning materials mentioned above, the categories of the aforementioned anomalies were used as the aforementioned labels and associated with the aforementioned thermal images used for learning.

45. As in request item 43, the exception handling procedure, wherein, The above-mentioned anomaly determination program causes the above-mentioned anomaly determination device to perform the step of generating the above-mentioned determination model based on the above-mentioned multiple learning materials.

46. ​​As in request item 44, the exception handling procedure, wherein, The above-mentioned anomaly determination program causes the above-mentioned anomaly determination device to perform the step of generating the above-mentioned determination model based on the above-mentioned multiple learning materials.

47. An exception handling procedure as described in any of requests 42 to 46, wherein, The above-mentioned anomalies include those related to the silk path.

48. An exception handling procedure as described in any of requests 43 to 46, wherein, The aforementioned anomaly determination program causes the aforementioned anomaly determination device to perform the step of acquiring multiple determination models learned according to the operating conditions of similar synthetic fiber manufacturing equipment. The determination step includes: selecting a determination model from the multiple determination models that corresponds to the current operating conditions of the synthetic fiber manufacturing equipment; and determining whether there is an anomaly based on the selected determination model and the thermal image used for determination input in the aforementioned input step.

49. An exception handling procedure as described in any of requests 42 to 46, wherein, The aforementioned synthetic fiber manufacturing equipment further includes an imaging device configured to capture images of the aforementioned guide fiber, and to acquire the aforementioned thermal image input during the aforementioned input step from the imaging device.

50. As in the exception handling procedure of request item 47, wherein, The aforementioned synthetic fiber manufacturing equipment further includes an imaging device configured to capture images of the aforementioned guide fiber, and to acquire the aforementioned thermal image input during the aforementioned input step from the imaging device.

51. As in the exception handling procedure of request item 49, where, The aforementioned anomaly determination program causes the aforementioned anomaly determination device to further execute the step of acquiring multiple determination models obtained by learning the positional relationship between the guide wire and the imaging device of the aforementioned synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment. The determination step includes: selecting a determination model from the multiple determination models that corresponds to the positional relationship between the guide wire and the imaging device of the aforementioned synthetic fiber manufacturing equipment; and determining whether there is an anomaly based on the selected determination model and the thermal image for determination input in the aforementioned input step.

52. As in the exception handling procedure of request item 50, where, The aforementioned anomaly determination program causes the aforementioned anomaly determination device to further execute the step of acquiring multiple determination models obtained by learning the positional relationship between the guide wire and the imaging device of the aforementioned synthetic fiber manufacturing equipment or other synthetic fiber manufacturing equipment. The determination step includes: selecting a determination model from the multiple determination models that corresponds to the positional relationship between the guide wire and the imaging device of the aforementioned synthetic fiber manufacturing equipment; and determining whether there is an anomaly based on the selected determination model and the thermal image for determination input in the aforementioned input step.

Images