Method, device and computer readable recording medium for generating a learned model

Abstract

This invention provides a method, apparatus, and computer-readable recording medium for generating a learned model. The device status diagnostic model generation unit (124) generates a shape data group corresponding to any one of multiple cores (206) and any one of multiple groups of data, and generates a swapped data group that swaps the correspondence between the cores (206) and the groups of data in all combinations of the multiple cores (206) and the multiple groups of data. This swapped data group is used to generate or update multiple learned models (M) whose representation defects are caused by any one of the multiple cores (206).

Description

Technical Field

[0001] This disclosure relates to a method, apparatus, and computer-readable recording medium for generating a learned model for displaying information related to the maintenance of production equipment. Background Technology

[0002] To prevent deterioration and malfunctions and maintain normal operation of a particular piece of equipment, a maintenance system is typically installed. Patent Document 1 discloses a maintenance system that monitors for anomalies such as drainage pump malfunctions and switchboard grounding in a substation. In the event of an anomaly, the system notifies relevant personnel and stores information related to the maintenance work performed by the notified personnel.

[0003] Prior art literature

[0004] Patent documents

[0005] Patent Document 1: Japanese Patent Application Publication No. 2017-167708 Summary of the Invention

[0006] One aspect of this disclosure relates to a learning-completed model generation method for maintaining a winding device, the winding device comprising: a first supply mechanism for supplying a first electrode sheet; a second supply mechanism for supplying a second electrode sheet; a first bonding roller disposed on the side of the first electrode sheet; a second bonding roller disposed on the side of the second electrode sheet and paired with the first bonding roller to bond the first electrode sheet and the second electrode sheet; a first winding core; a second winding core; and a driving mechanism for moving the first winding core to a given winding position, thereby winding the first electrode sheet into a winding position. The first electrode sheet and the second electrode sheet are overlapped and wound around the first core, and the second core is moved to the given winding position, so that the first electrode sheet and the second electrode sheet are overlapped and wound around the second core; and the sensor reads the first end face of the first electrode sheet and the second end face of the second electrode sheet along the radial direction of the first winding body on the first core where the first electrode sheet and the second electrode sheet are wrapped in multiple overlapping turns, and reads the first end face of the first electrode sheet and the second end face of the second electrode sheet along the radial direction of the first winding body on the second core where the first electrode sheet and the second electrode sheet are wrapped in multiple overlapping turns. The method for generating the learned model includes the following steps: acquiring from the sensor a first set of data representing the position of the first end face read along the radial direction of the second winding of the first electrode, a second set of data representing the position of the second end face read along the radial direction of the first winding, a third set of data representing the position of the third end face read along the radial direction of the second winding, and a fourth set of data representing the position of the fourth end face read along the radial direction of the second winding; based on the continuity of the position of the third end face represented by the third set of data, the continuity of the position of the fourth end face represented by the fourth set of data, and the positional relationship of the baseline, when it is determined that the second winding is defective, using the third set of data and the fourth set of data to generate a first learned model for outputting information indicating that the cause of the defect is the second winding core;A second learned model is generated using the first set of data, the second set of data, the third set of data, and the fourth set of data. This second learned model takes the fifth set of data and the sixth set of data as input, and outputs information indicating that the third winding is defective and the cause of the defect is the first core. The fifth set of data, representing the position of the fifth end face of the first electrode, is obtained from the sensor along the radial direction of the third winding, which involves multiple overlapping turns of the first and second electrode sheets on the first core. The sixth set of data, representing the position of the sixth end face of the second electrode, is obtained along the radial direction of the third winding. Based on the continuity of the position of the fifth end face indicated by the fifth set of data, the continuity of the position of the sixth end face indicated by the sixth set of data, and the positional relationship of the baseline, the third winding is determined to be defective.

[0007] One aspect of this disclosure relates to an information output device for displaying information related to the maintenance of a winding device, the winding device comprising: a first supply mechanism for supplying a first electrode sheet; a second supply mechanism for supplying a second electrode sheet; a first bonding roller disposed on the side of the first electrode sheet; a second bonding roller disposed on the side of the second electrode sheet and paired with the first bonding roller to bond the first electrode sheet and the second electrode sheet; a first core; a second core; a drive mechanism for moving the first core to a given winding position to overlap and wind the first electrode sheet and the second electrode sheet around the first core, and for moving the second core to the given winding position to overlap and wind the first electrode sheet and the second electrode sheet around the second core; and a sensor that, along the first core, has the first electrode sheet wound in multiple overlapping turns to overlap and wind the first electrode sheet. The device for outputting information includes: an acquisition unit that acquires from the sensor a first set of data indicating the position of the first end face read along the radial direction of the first winding of the second electrode sheet and the second end face of the first electrode sheet and the second end face of the second electrode sheet, and reads along the radial direction of the second winding of the second core in which the first electrode sheet and the second electrode sheet are wound in multiple overlapping turns, and a third set of data indicating the position of the third end face read along the radial direction of the second winding of the second core and the fourth end face of the second electrode sheet.The model generation unit, based on the continuity of the position of the third end face represented by the third set of data, the continuity of the position of the fourth end face represented by the fourth set of data, and the positional relationship of the baseline, generates a first learned model using the third set of data and the fourth set of data when it is determined that the second winding is defective. This model outputs information indicating that the defect is caused by the second winding core. A second learned model is generated using the first set of data, the second set of data, the third set of data, and the fourth set of data. This second learned model takes the fifth set of data and the sixth set of data as input in the following cases, and outputs information indicating that the third winding is defective. The reason for the defect is that the information of the first winding core is output. In the above case, the sensor acquires the fifth set of data indicating the position of the fifth end face of the first electrode sheet along the radial direction of the third winding body (which has multiple overlapping turns of the first and second electrode sheets on the first winding core), and the sixth set of data indicating the position of the sixth end face of the second electrode sheet along the radial direction of the third winding body. Based on the continuity of the position of the fifth end face indicated by the fifth set of data, the continuity of the position of the sixth end face indicated by the sixth set of data, and the positional relationship of the baseline, the third winding body is determined to be defective.

[0008] One aspect of this disclosure relates to a computer-readable recording medium that records a computer-executed program to generate a learned model for the maintenance of a winding device, the winding device comprising: a first supply mechanism for supplying a first electrode sheet; a second supply mechanism for supplying a second electrode sheet; a first bonding roller disposed on the side of the first electrode sheet; a second bonding roller disposed on the side of the second electrode sheet and paired with the first bonding roller to bond the first electrode sheet and the second electrode sheet; a first core; a second core; a drive mechanism for moving the first core to a given winding position to overlap and wind the first electrode sheet and the second electrode sheet around the first core, and for moving the second core to the given winding position to overlap and wind the first electrode sheet and the second electrode sheet around the second core; and a sensor that is wound in multiple overlapping turns around the first core. The program reads the first end face of the first electrode sheet and the second end face of the second electrode sheet along the radial direction of the first winding body of the first electrode sheet and the second winding body of the second winding body, which overlaps the first electrode sheet and the second electrode sheet in multiple turns on the second winding body. The program causes the computer to perform the following process: acquire from the sensor a first set of data indicating the position of the first end face read along the radial direction of the first winding body, a second set of data indicating the position of the second end face read along the radial direction of the first winding body, a third set of data indicating the position of the third end face read along the radial direction of the second winding body, and a fourth set of data indicating the position of the fourth end face read along the radial direction of the second winding body;Based on the continuity of the position of the third end face represented by the third set of data, the continuity of the position of the fourth end face represented by the fourth set of data, and the positional relationship of the baseline, when it is determined that the second winding is defective, a first learned model is generated using the third set of data and the fourth set of data to output information indicating that the cause of the defect is the second winding core. A second learned model is generated using the first set of data, the second set of data, the third set of data, and the fourth set of data. This second learned model takes the fifth set of data and the sixth set of data in the following cases as input, and takes the information indicating that the third winding is defective and the... The reason for the defect is the information meaning of the first winding core. In the above situation, the sensor acquires the fifth set of data indicating the position of the fifth end face of the first electrode sheet along the radial direction of the third winding body (which has multiple overlapping turns of the first and second electrode sheets on the first winding core), and the sixth set of data indicating the position of the sixth end face of the second electrode sheet along the radial direction of the third winding body. Based on the continuity of the position of the fifth end face indicated by the fifth set of data, the continuity of the position of the sixth end face indicated by the sixth set of data, and the positional relationship of the baseline, the third winding body is determined to be defective. Attached Figure Description

[0009] Figure 1 It is a network diagram that includes a maintenance display device and a winding device for applying the maintenance display device.

[0010] Figure 2 This is a flowchart illustrating the overall processing steps for maintaining a display device.

[0011] Figure 3A This is a diagram illustrating the structure of the winding section in a winding device for producing a wound body.

[0012] Figure 3B This is a perspective view illustrating a wound body produced in the winding section.

[0013] Figure 4A This is a schematic diagram illustrating the inspection process of a wound object by an inspection machine.

[0014] Figure 4B This is a schematic diagram illustrating the cross-sectional shape along the radial direction of the wound body.

[0015] Figure 4C This illustrates the scanning process of the inspection machine. Figure 4B The image is a cross-section of the coil shown.

[0016] Figure 5This is a schematic diagram illustrating an example of the cross-sectional shape and shape data of a winding when defects occur due to windings wound on different cores.

[0017] Figure 6 This is a block diagram illustrating the functional structure of the maintenance display device according to the first embodiment.

[0018] Figure 7A This is a graph that illustrates actual production results data.

[0019] Figure 7B This is a graph that illustrates actual production results data.

[0020] Figure 8 This is a diagram illustrating the maintenance of actual results data.

[0021] Figure 9 It is a timing diagram that roughly illustrates the overall process flow of the maintenance display device.

[0022] Figure 10 It is a timing diagram that roughly illustrates the overall process flow of the maintenance display device.

[0023] Figure 11 This is a flowchart used to explain the processes performed by the maintenance effect judgment unit in the learning process.

[0024] Figure 12A It is a conceptual diagram used to illustrate the pattern of judging the effectiveness of maintenance operations in learning processing.

[0025] Figure 12B It is a conceptual diagram used to illustrate the pattern of judging the effectiveness of maintenance operations in learning processing.

[0026] Figure 13 This is a flowchart illustrating the processes performed by the device status diagnostic model generation unit during the learning process.

[0027] Figure 14 This is a diagram used to illustrate the swapping of data groups.

[0028] Figure 15 This is a flowchart illustrating the processes performed by the device status diagnostic unit during the identification process.

[0029] Figure 16 This is a flowchart used to explain the process performed by the decision-making unit in the identification process.

[0030] Figure 17A This is a diagram illustrating a specific example of maintenance team information.

[0031] Figure 17BThis is a diagram showing specific examples of a list of maintenance plans.

[0032] Figure 18 This is a flowchart used to explain the processes performed by the maintenance effect determination unit during the update process.

[0033] Figure 19A It is a conceptual diagram used to illustrate the pattern of judging the effectiveness of maintenance operations in the update process.

[0034] Figure 19B It is a conceptual diagram used to illustrate the pattern of judging the effectiveness of maintenance operations in the update process.

[0035] Figure 20 This is a flowchart illustrating the processes performed by the device status diagnostic model generation unit during the update process.

[0036] Figure 21 This is a diagram illustrating the structure of the maintenance display device according to the second embodiment.

[0037] Figure 22 This is a flowchart for explaining the processing performed by the maintenance effect determination unit in the second embodiment.

[0038] Figure 23 This is a diagram illustrating the structure of the maintenance display device according to the third embodiment.

[0039] Figure 24 This is a flowchart explaining the processing performed by the device status diagnostic model generation unit in the third embodiment.

[0040] Figure 25 This is a flowchart for explaining the processing performed by the notification determination unit in the third embodiment.

[0041] Figure 26A This is a diagram illustrating a variation of the method for determining the effectiveness of maintenance work performed by the maintenance effectiveness determination unit in the learning process.

[0042] Figure 26B This is a diagram illustrating a variation of the method for determining the effectiveness of maintenance work performed by the maintenance effectiveness determination unit in the learning process.

[0043] Figure 27A This is a diagram illustrating a variation of the method for determining the effectiveness of maintenance operations performed by the maintenance effectiveness determination unit during the update process.

[0044] Figure 27B This is a diagram illustrating a variation of the method for determining the effectiveness of maintenance operations performed by the maintenance effectiveness determination unit during the update process.

[0045] Symbol Explanation

[0046] Servers 10, 10A, and 10B;

[0047] 50 First supply reels;

[0048] 51. Second supply reel;

[0049] 100, 100A, 100B Maintenance display devices;

[0050] Storage units 110 and 110B;

[0051] 111 Database of Actual Production Results;

[0052] 112 Equipment Status Diagnostic Model Database;

[0053] 113 Maintain the database of actual results;

[0054] 114 Database of ineffective equipment status diagnostic models;

[0055] 120, 120A, 120B Control Unit;

[0056] 121 Equipment Status Diagnostics Department;

[0057] Notify the Judgment Department of 122 and 122B;

[0058] Maintenance effectiveness assessment department (123, 123A, 123B);

[0059] Equipment status diagnostic model generation unit (124, 124B);

[0060] 130 Notification Department;

[0061] 131 Alarm Department;

[0062] 132 Display Unit;

[0063] 200 winding device;

[0064] 201 Winding section;

[0065] 202 Sheet No. 1;

[0066] 203 Second sheet;

[0067] 204, 204α, 204β, 204γ wound bodies;

[0068] 205A First Laminating Roller;

[0069] 205B Second Laminating Roller;

[0070] 206, 206α, 206β, 206γ core;

[0071] 206M core rotation drive unit;

[0072] 207 Inspection Machine;

[0073] 208 turntables;

[0074] 209 Cut-off section;

[0075] 210 Pressing part;

[0076] 211 Joint welding section;

[0077] 212 With adhesive part;

[0078] 213 Roller. Detailed Implementation

[0079] In the technology disclosed in Patent Document 1, personnel related to the equipment are notified after an equipment malfunction occurs. Therefore, maintenance performed by these personnel occurs after the malfunction has occurred. Since maintenance is performed after an malfunction, the equipment needs to be stopped, it is desirable to notify the relevant personnel before the malfunction occurs, at a point when maintenance is deemed necessary. Therefore, it is necessary to detect any signs of an impending equipment malfunction.

[0080] The purpose of this disclosure is to provide a method, apparatus, and computer-readable recording medium for generating a learned model for detecting signs of anomalies.

[0081] Hereinafter, various embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. However, sometimes the necessary detailed descriptions are omitted, for example, detailed descriptions of matters that are already well known, or repeated descriptions of substantially the same structures.

[0082] Furthermore, the following description and accompanying drawings are provided to enable those skilled in the art to understand this disclosure, and are not intended to limit the technical solutions of this disclosure.

[0083] (First embodiment)

[0084] <Maintenance display device 100 and winding device 200>

[0085] Figure 1 This is a network diagram including the maintenance display device 100 according to the first embodiment of this disclosure and the winding device 200 using the maintenance display device 100. The maintenance display device 100 described in this embodiment is an apparatus for displaying maintenance information for the winding device 200 used in the production of lithium-ion secondary batteries. Figure 1In the example shown, the maintenance display device 100 is applied to one winding device 200, but this disclosure is not limited to this, and a maintenance display device may also be applied to multiple winding devices. Furthermore, in this embodiment, the maintenance display device 100 is described as an apparatus, but this disclosure is not limited to this, and it may also be a maintenance display system in which various structures are connected via a network.

[0086] The maintenance display device 100 includes: a server 10, which has a storage unit 110 and a control unit 120; and a notification unit 130. The server 10 is communicatively connected to the scrolling device 200 via a network NT. The network NT is, for example, a public network such as the Internet, or a local area network such as a company's internal LAN (Local Area Network).

[0087] Server 10 is, for example, a general-purpose computer, such as... Figure 1 As shown, it has a storage unit 110 and a control unit 120.

[0088] Storage unit 110 includes, for example, main storage devices such as ROM (Read Only Memory) and RAM (Random Access Memory) (not shown), and / or auxiliary storage devices such as HDD (Hard Disk Drive), SSD (Solid State Drive), and flash memory (not shown).

[0089] The control unit 120 is a hardware processor (not shown), such as a CPU (Central Processing Unit), which expands and executes the program stored in the storage unit 110, thereby controlling the overall maintenance display device 100.

[0090] The storage unit 110 and the control unit 120 may not be configured as a single computer. That is, the storage unit 110 and the control unit 120 can be configured to communicate with each other, or they can be configured as separate units located in separate positions. Furthermore, the maintenance display device 100 may also have... Figure 1 The operation unit (not shown) receives operation input from external sources. Details regarding the storage unit 110 and the control unit 120 will be described later.

[0091] exist Figure 1In the example shown, the notification unit 130 is included in the winding device 200 and is connected to the server 10 via the network NT. The notification unit 130 notifies the user maintaining the display device 100 based on the control of the control unit 120. Furthermore, in this embodiment, the term "user maintaining the display device 100" includes the administrator of the display device 100, or the person using the winding device 200 to wind the device (see below). Figure 3B Production workers, etc.

[0092] like Figure 1 As shown, the notification unit 130 includes an alarm unit 131 and a display unit 132. The alarm unit 131 is, for example, a buzzer, a light, or a similar device that alerts the user through sound or light. The display unit 132 is, for example, a liquid crystal display (LCD), an organic EL display, or a similar display device, and is used to display the warning content. Alternatively, in addition to the alarm unit 131 and the display unit 132, the notification unit 130 may also include a sending unit that sends emails containing warning content to, for example, pre-registered user email addresses.

[0093] In this embodiment, the winding device 200 is an apparatus for winding positive and negative electrode sheets to produce lithium-ion secondary batteries. Figure 1 As shown, the winding apparatus 200 includes a winding section 201 and an inspection machine 207. Details of the winding section 201 will be described later; it winds up positive and negative electrode sheets to produce a wound body. The inspection machine 207 inspects the wound body produced by the winding section 201.

[0094] In addition, Figure 1 In the example shown, the notification unit 130 is included in the winding device 200, but this disclosure is not limited thereto; the notification unit 130 may also be provided outside the winding device 200. Furthermore, in Figure 1 In the example shown, the notification unit 130 is connected to the server 10 via the network NT, but this disclosure is not limited to this, and the server 10 and the notification unit 130 may also be directly connected without going through the network NT.

[0095] Furthermore, in this embodiment, the winding device 200 is described as a winding device for winding the positive and negative electrode sheets of a lithium-ion secondary battery, but this disclosure is not limited to this. The maintenance display device of this disclosure can also be applied to production equipment other than lithium-ion secondary battery winding devices. Moreover, the maintenance display device of this disclosure can also be applied to various other types of equipment besides production equipment.

[0096] Figure 2 This is a flowchart illustrating the overall processing steps for maintaining the display device 100.

[0097] In step S1, the control unit 120 causes the winding unit 201 of the winding device 200 to produce a wound body.

[0098] In step S2, the control unit 120 causes the inspection machine 207 to inspect the produced wound body. Details regarding the inspection of the wound body performed by the inspection machine 207 will be described later.

[0099] In step S3, the control unit 120 causes the storage unit 110 to store the inspection results based on the inspection machine 207. Simultaneously, in step S4, the control unit 120 determines whether the wound body is a defective product based on the inspection results from the inspection machine 207. If it is determined that the product is not defective (step S4: No), the control unit 120 proceeds the process to step S5; if it is determined that the product is defective (step S4: Yes), the process proceeds to step S6.

[0100] If the product is determined not to be defective, in step S5, the control unit 120 causes the winding device 200 to supply the winding body to the next process.

[0101] If the product is determined to be defective, in step S6, the control unit 120 notifies the notification unit 130 that a defective product has been detected. Details regarding the notification executed by the notification unit 130 will be described later.

[0102] In step S7, the control unit 120 causes the winding device 200 to discard the winding that is determined to be defective.

[0103] In addition, Figure 2 In steps S5 and S7 of the flowchart shown, the control unit 120 causes the winding device 200 to supply the winding body to the next process or discard the winding body, but this disclosure is not limited to this. For example, the notification unit 130 can also notify the user of the maintenance display device 100 to supply the winding body to the next process or discard the winding body, thereby causing the user to supply or discard the winding body.

[0104] The winding section 201 of the winding device 200 and the inspection machine 207 will now be described in detail.

[0105] <Wrapping section 201>

[0106] Figure 3A This is a diagram illustrating the structure of the winding section 201.

[0107] like Figure 3AAs shown, the winding unit 201 includes a first supply reel 50, a second supply reel 51, a first bonding roller 205A, a second bonding roller 205B, a core 206 (206α, 206β, 206γ), a core rotation drive unit 206M, a turntable (index table) 208, a cutting unit 209, a pressing unit 210, a joint welding unit 211, a tape bonding unit 212, and a cylinder (cylinder) 213. The winding unit 201 is an apparatus that uses the first bonding roller 205A and the second bonding roller 205B to bond and wind a first sheet 202 supplied from the first supply reel 50 and a second sheet 203 supplied from the second supply reel 51 onto the core 206 to produce a wound body 204. The core rotation drive unit 206M drives the core 206 at a desired rotational speed.

[0108] The first sheet 202 is, for example, a sheet-shaped component coated with a positive electrode material (positive electrode sheet), and the second sheet 203 is, for example, a sheet-shaped component coated with a negative electrode material (negative electrode sheet). The first sheet 202 is an example of the first electrode sheet of this disclosure, and the second sheet 203 is an example of the second electrode sheet of this disclosure. Furthermore, in the above examples, the first sheet 202 is a positive electrode sheet and the second sheet 203 is a negative electrode sheet, but this disclosure is not limited to this; it is also possible for the first sheet 202 to be a negative electrode sheet and the second sheet 203 to be a positive electrode sheet.

[0109] exist Figure 3A In the example shown, turntable 208 holds three cores 206α, 206β, and 206γ. Any one of these three cores 206α, 206β, and 206γ is an example of the second core of this disclosure, and the others are examples of the first core of this disclosure. In the following description, the three cores 206α, 206β, and 206γ are sometimes referred to collectively as core 206.

[0110] The turntable 208 rotates gradually at given intervals, causing the cores 206 to rotate along circular tracks. Thus, one of the three cores 206 is positioned in the take-up position. The take-up position is the position where the core 206 can be rotated by the core rotation drive unit 206M. Figure 3A In the example shown, core 206α is positioned at the take-up position. Once the take-up of one core 206 is complete, the turntable 208 sequentially switches to the next core 206. In the following description, the winding body taken up to core 206α is referred to as winding body 204α, the winding body taken up to core 206β is referred to as winding body 204β, and the winding body taken up to core 206γ is referred to as winding body 204γ. Additionally, in the following description, sometimes the three winding bodies 204α, 204β, and 204γ are collectively referred to as winding body 204.

[0111] In addition, Figure 3A In the example shown, the structure of the turntable 208 switching three cores in sequence is illustrated, but this disclosure is not limited to this. The number of cores 206 held by the turntable 208 can be two or more, or even several.

[0112] When the cutting section 209 completes winding in a core 206, it cuts the first sheet 202 and the second sheet 203. At this time, the pressing section 210 presses the winding body 204 wound on the core 206 and presses down on the vibration at the ends of the cut first sheet 202 and second sheet 203. Figure 3A In the example shown, the cutting part 209 is positioned to cut before the first sheet 202 and the second sheet 203 are bonded together, but it can also be positioned to cut after the first sheet 202 and the second sheet 203 are bonded together.

[0113] The joint welding section 211 welds the current collector to the first sheet 202. When the tape bonding section 212 is cut by the cutting section 209 after winding in the core 206, it is secured with tape to prevent the wound body 204 from vibrating. The roller 213 adjusts the tension applied to the first sheet 202 and the second sheet 203 via the second bonding roller 205B.

[0114] Figure 3B This is a perspective view illustrating a wound body 204 produced in the winding section 201. Figure 3B The image shows the state where the terminal portions (the ends cut off by the cutting portion 209) of the first sheet 202 and the second sheet 203 constituting the winding body 204 are not wound. Figure 3B As shown, the width (length along the axial direction of the winding 204) of the second sheet 203 is larger than that of the first sheet 202.

[0115] <Inspection Machine 207>

[0116] Inspection machine 207 inspects the produced wound body 204. Inspection machine 207 is, for example, an SS-OCT (SweptSource-Optical Coherence Tomography) device. Inspection machine 207 is an example of the sensor disclosed herein.

[0117] Figure 4A This is a schematic diagram illustrating the state of the inspection machine 207 inspecting the wound body 204. For example... Figure 4A As shown, the inspection machine 207 scans the wound body 204, which is illuminated by light L, by moving the light L from the radial inside to the outside of the wound body 204, and uses the coherence of the light L to generate an image representing the shape of the internal structure of the wound body 204.

[0118] Figure 4B This is a schematic diagram illustrating the cross-sectional shape along the radial direction of the wound body 204. Furthermore, Figure 4C This example illustrates the inspection machine 207 pairs. Figure 4B Image I is a diagram generated by scanning a cross-section of the shown wound body 204. Figure 4B as well as Figure 4C In the middle, the vertical direction corresponds to the axial direction of the winding body 204, and the horizontal direction corresponds to the radial direction of the winding body 204.

[0119] like Figure 4B As shown, in a cross-section along the radial direction of the wound body 204, a first sheet 202 and a second sheet 203, which is wider than the first sheet 202, are alternately stacked. An inspection machine 207 extracts and images the positions of the two ends of the wound body 204 along the axial direction of the first sheet 202 and the two ends along the axial direction of the second sheet 203 in the radial direction. Figure 4C In the example shown (Image I), the rhombus α corresponds to the first sheet end position data group, representing the positions of the two ends of the first sheet 202, and the black circle β corresponds to the second sheet end position data group, representing the positions of the two ends of the second sheet 203. The first sheet end position data group is an example of the first, third, or fifth data group of this disclosure. The second sheet end position data group is an example of the second, fourth, or sixth data group of this disclosure.

[0120] During the production of the wound body 204 in the winding section 201, defective products may sometimes be produced. For example, defective products may be produced due to the aforementioned defects in the various structures of the winding section 201. The inspection machine 207 generates an image showing the cross-sectional shape along the radial direction of the wound body 204 as described above, and stores it as shape data in the storage unit 110. The result of determining whether the wound body 204 is defective based on the shape data is also stored in the storage unit 110. In addition, the determination of whether a product is defective based on the shape data can be performed by... Figure 1 The control unit 120 shown can perform the operation, or it can be performed by the inspection machine 207, or it can be performed by... Figure 1 or Figure 3A Other structures not shown in the diagram are also included. The following describes the process by which the control unit 120 determines whether a product is defective based on shape data.

[0121] Figure 5 This is a schematic diagram showing an example of the cross-sectional shape and shape data of the wound body 204 under the condition that defects occur when the wound body 204 is wound on different cores 206. Figure 5 The upper part, for example, is in Figure 3AIn each of the shown cores 206α, 206β, and 206γ, in Figure 3A The cross-sectional shapes of the wound bodies 204α, 204β, and 204γ respectively, formed during one rotation of the turntable 208. Figure 5 In the example shown, in the windings 204α and 204β, the heights of the two ends of the first sheet 202 and the second sheet 203 are the same, but in the winding 204γ, the heights of the two ends of the first sheet 202 and the second sheet 203 become inclined. Furthermore, one rotation of the turntable 208 means that all the cores 206α, 206β, and 206γ are approximately positioned at the winding position and the winding 204 is wound up at each core 206.

[0122] exist Figure 5 The lower part illustrates examples based on... Figure 5 Images Iα, Iβ, and Iγ are generated based on the cross-sectional shapes of the upper portions of the coiled bodies 204α, 204β, and 204γ. In the following description, they will be compared with... Figure 3A The images Iα, Iβ, and Iγ corresponding to the cores 206α, 206β, and 206γ shown are collectively recorded as a shape data group. That is, a shape data group represents... Figure 3A The shape data (images) are a set of cross-sectional shapes of the wound bodies 204α, 204β, 204γ respectively wound on the cores 206α, 206β, 206γ during one rotation of the turntable 208 shown. In other words, the shape data set refers to a set of shape data corresponding to the individual cores of the multiple cores 206.

[0123] Furthermore, the correspondence between the core and the image in the shape data group is recorded as a dataset. For example, an image Iα generated based on the shape data of the core 206α and the winding body 204α wound around the core 206α is given as a corresponding dataset DSα. The shape data group includes datasets DSα, DSβ, and DSγ (see below). Figure 14 ).

[0124] So-called Figure 5 The reference lines shown are lines that indicate the proper positions of the two ends of the first sheet 202 and the second sheet 203, serving as references. For example... Figure 5 As shown, in images Iα and Iβ, the positions of the two ends of the first sheet 202 and the second sheet 203 are continuous with the baseline (parallel to the baseline), but in image Iγ, the positions of the two ends of the first sheet 202 and the second sheet 203 become inclined from the baseline.

[0125] like Figure 5As shown in the winding bodies 204α and 204β, a winding body with a continuous cross-sectional shape parallel to the reference line at the positions of both ends of the first sheet 202 and the second sheet 203 is judged as having an inspection result of "good" by the inspection machine 207. On the other hand, as Figure 5 As shown in the winding body 204γ, a winding body with a cross-sectional shape that is continuously inclined relative to the baseline at the positions of the two ends of the first sheet 202 and the second sheet 203 is judged as "defective" by the inspection machine 207.

[0126] Thus, the defect of the continuous tilting of the two ends of the first sheet 202 and the second sheet 203 from the baseline is mainly likely to occur when the core 206 has a defect. A defect in the core 206 is, for example, a condition where one side of the core 206 is cut due to wear or other reasons. It can be assumed that if any one of the multiple cores 206 has a defect, only the winding body 204 wound around the defective core 206 is considered defective.

[0127] In addition, Figure 5 In the example shown, the quality of the wound body 204 is determined to be "good" or "bad" based on whether the continuity of the positions of the two ends of the first sheet 202 and the second sheet 203 is parallel to the reference line. More specifically in this disclosure, even when the continuity of the positions of the two ends of the first sheet 202 and the second sheet 203 becomes inclined from the reference line, the degree of defect can be graded based on the magnitude of the inclination. Specifically, for example, if the inclination angle of the continuity of the positions of the two ends of the first sheet 202 and the second sheet 203 is below a given value, the wound body 204 is judged to be "passable," and if the inclination angle exceeds the given value, the wound body 204 is judged to be "bad." Furthermore, the above-described method for judging defects is just one example, and there is no particular limitation in this disclosure regarding what constitutes a defect.

[0128] <Maintenance Display Device 100>

[0129] The functional structure and operation of the maintenance display device 100, which displays information related to maintenance work to be performed on the aforementioned winding device 200, will now be described in detail. Furthermore, maintenance work in this embodiment means that by appropriately adjusting various structures and replacing parts of the winding device 200, defects are prevented from occurring in the wound body 204 produced by the winding device 200. In this disclosure, maintenance work specifically refers to work performed to maintain the aforementioned core 206 in good condition. Maintenance work is performed by personnel who actually operate the winding device 200.

[0130] <Storage Department 110>

[0131] Figure 6 This is a block diagram illustrating the functional structure of the maintenance display device 100 according to the first embodiment. As described above, the maintenance display device 100 includes a storage unit 110, a control unit 120, and a notification unit 130 (see reference 120). Figure 1 ).

[0132] like Figure 6 As shown, the storage unit 110 has a production actual results database 111, an equipment status diagnosis model database 112, and a maintenance actual results database 113.

[0133] The production results database 111 is a database that records production results data related to the production results of the winding device 200. The production results data includes the production date and time of the produced winding 204 and the shape data of the winding 204.

[0134] Figure 7A as well as Figure 7B This is a graph illustrating actual production results data (PD). Figure 7A In the table, a portion of the actual production results (PD) data is presented in tabular form. For example... Figure 7A As shown, the actual production results data (PD) includes various data such as "production date and time", "equipment", "inspection results", "sheet 1", "sheet 2", and "shape data group ID".

[0135] The "Production Date and Time" data relates to the production date and time of the wound body 204. The "Equipment" data is used to identify the equipment used to provide the actual production results when multiple winding devices 200 are present. Figure 7A As an example, the identifiers “A”, “B”, and “C” of the different winding devices 200 are shown.

[0136] The "Inspection Results" data represents the inspection results of the wound body 204 produced in the winding device 200 (see reference). Figure 5 (Data). In Figure 7A In the example, "good" or "bad" is shown as the inspection result.

[0137] The "Sheet 1" data and the "Sheet 2" data are data related to the materials used in the production of the winding 204. As the "Sheet 1" data and the "Sheet 2" data, identifiers for identifying the respective materials are stored.

[0138] The "Shape Data Group ID" is an identification number established corresponding to the shape data groups containing datasets DSα, DSβ, and DSγ. Figure 7B The example illustrates the correspondence between shape data group IDs and shape data.

[0139] In this production output data (PD), all data except for the shape data group, such as each time a wound body 204 is produced in the winding device 200, are automatically or manually entered and registered in the production output database 111. The shape data is then inspected by the production wound body 204 by the inspection machine 207 (see reference). Figure 1 or Figure 4A The data is generated during inspection and registered in correspondence with the shape data ID. That is, the actual production result data PD essentially contains the shape data of the wound body 204. Therefore, in the actual production result database 111, whenever the wound body 204 is produced, the actual production result data PD of the produced wound body 204 is registered.

[0140] The equipment condition diagnostic model database 112 is a database that registers multiple equipment condition diagnostic models M. Each equipment condition diagnostic model M is a learned model used as a diagnostic benchmark to determine whether maintenance work is required on the winding device 200. The equipment condition diagnostic model M is a learned model that has learned which maintenance work is effective for which type of defect when the winding device 200, which produces defective products, is improved through maintenance work (the production ratio of defective products decreases). More specifically, the equipment condition diagnostic model M is a collection of data including the shape data of a winding containing multiple defective products and the content of maintenance work performed to improve the defect of those defective products. The equipment condition diagnostic model M is generated by the equipment condition diagnostic model generation unit 124, described later.

[0141] The equipment condition diagnostic model M is generated whenever a maintenance operation is performed that reduces the proportion of defective products in the production of the wound body thereafter. That is, for example, the equipment condition diagnostic model M for yesterday's maintenance operation and the equipment condition diagnostic model M for today's maintenance operation are generated independently.

[0142] Furthermore, the form of the equipment condition diagnostic model M is not particularly limited, but in order to further improve diagnostic accuracy, it is desirable to use machine learning models such as neural network models. The selection of the model used in the equipment condition diagnostic model M can be made by the user of the maintenance display device 100 through an operation unit (not shown) or by the equipment condition diagnostic model generation unit 124.

[0143] The maintenance actual results database 113 is a database that registers maintenance actual results data MD related to the actual maintenance work performed on the winding device 200. The maintenance actual results data MD includes, for example, equipment data for identifying the winding device 200, data related to the date and time of the maintenance work (maintenance date and time), and data indicating the content of the maintenance work performed. For example, in the case of a short maintenance work that ends in a few minutes, the maintenance date and time can be either the start time or the end time of the maintenance work. Furthermore, in the case of a longer maintenance work, such as one that takes several hours, the maintenance date and time is preferably set to the midpoint of the maintenance work. Figure 8 This diagram illustrates the maintenance actual results data (MD). The maintenance actual results data (MD) is generated immediately after maintenance work is performed, for example, by personnel who have actually performed maintenance work on the winding device 200. Figure 1 Operational components (not shown) are input to the maintenance display device 100.

[0144] <Control Unit 120>

[0145] like Figure 6 As shown, the control unit 120 includes an equipment status diagnosis unit 121, a notification determination unit 122, a maintenance effect determination unit 123, and an equipment status diagnosis model generation unit 124.

[0146] The equipment condition diagnosis unit 121 diagnoses the condition of the winding device 200 using the shape data of the newly produced wound body 204 in the winding device 200 and the equipment condition diagnosis model M. The diagnosis result is calculated as a consistency degree C, which represents the degree of consistency between the shape data of the newly produced wound body 204 and the past shape data included in the equipment condition diagnosis model M. Here, the equipment condition diagnosis model M includes the content of maintenance operations and the shape data prior to the time point when the maintenance operation was performed. This means that in the past, when a defect occurred in the wound body 204 with the shape data included in the equipment condition diagnosis model M, the defect of the wound body 204 was reduced by performing the maintenance operations included in the equipment condition diagnosis model M. That is, the consistency degree C between the shape data of the newly produced wound body 204 and the shape data included in the equipment condition diagnosis model M represents the probability that the defect of the wound body 204 is improved by performing the maintenance included in the equipment condition diagnosis model M.

[0147] Furthermore, regarding the method of calculating the consistency degree C by comparing multiple (m) shape data of the newly produced winding body 204 with multiple (n) past shape data contained in the equipment condition diagnostic model M, appropriate methods include pattern matching or deep learning using feature quantities of multiple shape data that have undergone dimensionality compression. Alternatively, the consistency degree can also be calculated based on the distance between vectors obtained from each shape data.

[0148] The notification determination unit 122 determines whether to issue a notification related to maintenance work on the winding device 200 based on a consistency degree C. If the consistency degree C is above a given threshold, the notification determination unit 122 determines that a notification indicating that maintenance work should be performed should be issued; if the consistency degree C is below the given threshold, it determines that no notification should be issued. Notifications related to maintenance work include alarms to attract the user's attention, displays informing the user of the predictable effects of maintenance work performed.

[0149] The maintenance effect determination unit 123 determines whether the maintenance operation on the winding device 200 has any effect. The maintenance effect determination unit 123 may base its determination on, for example, the defect rate (the ratio of defective products to the total production) before and after the maintenance operation, or the shape data of the wound body 204 before and after the maintenance operation (see reference). Figure 5 This determines the effect of maintenance work.

[0150] The equipment condition diagnostic model generation unit 124 generates an equipment condition diagnostic model M based on the actual maintenance result data MD that is determined to be effective and the shape data of defective products produced before the maintenance operation. The equipment condition diagnostic model M generated by the equipment condition diagnostic model generation unit 124 is registered in the aforementioned equipment condition diagnostic model database 112.

[0151] <Overall Process Flow in Maintenance Display Device 100>

[0152] Below, refer to Figure 9 as well as Figure 10 For those with Figure 6 The overall process flow of the maintenance display device 100 with the shown functional structure will be explained. Figure 9 as well as Figure 10 This is a timing diagram that roughly illustrates the overall flow of the processing in the maintenance display device 100.

[0153] exist Figure 9 The diagram shows an outline of the learning process in the maintenance display device 100 and the recognition process utilizing the learned model generated by the learning process.

[0154] [Learning Processing]

[0155] The learning process in the maintenance display device 100 is a process used to generate a learned model (equipment condition diagnostic model M) of a defective product that has learned what shape data was produced by the winding device 200 and how it was improved through what maintenance operations. Therefore, the prerequisite for the learning process is that maintenance operations are performed before it begins.

[0156] In step S11, the maintenance effect determination unit 123 acquires the actual production result data PD of the multiple windings 204 produced before the start of the learning process (refer to...). Figure 7B The shape data contained in ) (refer to Figure 5 Based on this data, the failure rate Nf before maintenance work is calculated. before Defect rate Nf before For example, it is calculated by dividing the number of defective wound bodies 204 produced before the maintenance operation by the total number produced before the maintenance operation.

[0157] In step S12, the maintenance effect determination unit 123 acquires the shape data contained in the actual production result data PD of multiple wound bodies 204 produced after the maintenance operation, and calculates the defect rate Nf after the maintenance operation based on the data. after Defect rate Nf after For example, it is calculated by dividing the number of defective wound bodies 204 produced after maintenance by the total number of wound bodies 204 produced after maintenance.

[0158] In step S13, the maintenance effect determination unit 123 determines the defect rate Nf before and after the maintenance operation. before and f after The effectiveness of the maintenance work is determined by comparison. Details regarding the determination of the effectiveness of maintenance work performed by the maintenance effectiveness determination unit 123 during the learning process will be described later.

[0159] If the maintenance operation is determined to be effective in step S13, the maintenance effectiveness determination unit 123 will, in step S14, display the actual maintenance result data MD (referencing) that represents the content of the maintenance operation performed before the start of the learning process. Figure 8 The output is sent to the device status diagnostic model generation unit 124.

[0160] In step S15, the equipment condition diagnostic model generation unit 124 generates an equipment condition diagnostic model M using the maintenance actual result data MD that is determined to be effective. Details regarding the equipment condition diagnostic model M will be described later.

[0161] In step S16, the equipment condition diagnostic model generation unit 124 registers the generated equipment condition diagnostic model M into the equipment condition diagnostic model database 112 (see reference). Figure 6 ).

[0162] The steps S11 to S16 described above are the learning processes for maintaining the display device 100.

[0163] [Identification Processing]

[0164] The identification process described below is based on shape data representing the cross-sectional shape of multiple newly produced windings 204, and uses a device condition diagnostic model M generated through learning processing to identify whether an abnormality or a sign of an abnormality has occurred in the winding device 200.

[0165] In step S17, the equipment status diagnostic unit 121 acquires the shape data of the newly produced multiple wound bodies (hereinafter referred to as the new shape data).

[0166] In step S18, the equipment condition diagnosis unit 121 calculates the consistency degree C using the new shape data and the equipment condition diagnosis model M. The consistency degree C is a value that represents the degree of consistency between the new shape data and the past shape data contained in the equipment condition diagnosis model M. That is, the higher the consistency degree C, the higher the probability of an abnormality or a sign of an abnormality occurring in the winding device 200, and the higher the probability that the newly produced wound body 204 will become a defective product.

[0167] In step S19, the notification determination unit 122 determines that a notification is required to the user of the maintenance display device 100 if the consistency degree C is above a given threshold. A consistency degree C above the given threshold corresponds to a situation where an abnormality or signs of an abnormality occur in the winding device 200, necessitating the re-performing of maintenance work.

[0168] In step S110, the notification determination unit 122 outputs the maintenance operation content to be notified to the notification unit 130. The maintenance operation content to be notified to the user is determined based on the equipment condition diagnosis model M, where the consistency C is above a given threshold.

[0169] In steps S111 and S112, the notification unit 130 notifies the user that maintenance work should be performed. In step S111, the alarm unit 131 issues an alarm. Furthermore, in step S112, the display unit 132 displays the details of the maintenance work notified to the user. Figure 9 Although an example is shown of both issuing an alarm in step S111 and displaying the content of the maintenance operation in step S112, it is also possible, for example, to not issue an alarm and only display the content of the maintenance operation.

[0170] Thus, through the notifications in steps S111 and S112, the operator who receives the notification performs maintenance work on the winding device 200 based on the content of the notified maintenance work.

[0171] The processing of steps S17 to S112 described above utilizes the recognition processing of the maintenance display device 100 of the learned model generated through learning processing.

[0172] exist Figure 10The diagram shows an outline of the update process in the maintenance display device 100 and the recognition process utilizing the learned model updated through the update process.

[0173] [Update Processing]

[0174] The update process in the maintenance display device 100 is a process that updates the learned model (equipment condition diagnosis model M) based on the results of a new maintenance operation performed after the learning process described above. That is, the prerequisite for the update process is that a maintenance operation was performed before it began.

[0175] In step S21, the maintenance effect determination unit 123 uses the actual production result data PD of the multiple windings 204 produced before the new maintenance operation (refer to...) Figure 7A The shape data contained in ) (refer to Figure 5 The consistency C before maintenance work is calculated using the past shape data contained in the equipment condition diagnostic model M registered in the equipment condition diagnostic model database 112, as well as the data from the previous shape data. before .

[0176] In step S22, the maintenance effect determination unit 123 uses the shape data contained in the actual production results data of multiple windings 204 produced after the new maintenance operation, and the past shape data contained in the equipment condition diagnosis model M registered in the equipment condition diagnosis model database 112, to calculate the consistency C after the maintenance operation. after .

[0177] In step S23, the maintenance effect determination unit 123 determines the consistency C before and after the maintenance operation. before and C after The results are compared to determine whether the maintenance work was effective. Details regarding the determination of the effectiveness of maintenance work performed by the maintenance effectiveness determination unit 123 during the update process will be described later.

[0178] If the maintenance operation is deemed effective in step S23, the maintenance effect determination unit 123 outputs the actual maintenance result data MD, which represents the content of the maintenance operation performed before the update process begins, to the equipment condition diagnosis model generation unit 124 in step S24.

[0179] In step S25, the equipment condition diagnostic model generation unit 124 updates the equipment condition diagnostic model M using the maintenance actual result data MD that has been determined to be effective. Details regarding the update process of the equipment condition diagnostic model M will be described later.

[0180] In step S26, the equipment condition diagnosis model generation unit 124 uses the generated equipment condition diagnosis model M to update the equipment condition diagnosis model database 112 (see reference). Figure 6 ).

[0181] The steps S21 to S26 described above are for maintaining the update process of the display device 100.

[0182] [Identification Processing]

[0183] The identification process described below is based on shape data representing the cross-sectional shape of the multiple newly produced windings 204, and uses the equipment condition diagnostic model M updated through update processing to identify whether an abnormality or a sign of an abnormality has occurred in the winding device 200.

[0184] In step S27, the equipment status diagnostic unit 121 acquires the shape data of the newly produced multiple wound bodies (hereinafter referred to as the new shape data).

[0185] In step S28, the equipment condition diagnosis unit 121 calculates the consistency degree C using the new shape data and the equipment condition diagnosis model M. The consistency degree C is a value representing the degree of consistency between the new shape data and the past shape data contained in the equipment condition diagnosis model M.

[0186] In step S29, the notification determination unit 122 determines that the user of the maintenance display device 100 needs to be notified if the consistency degree C is above a given threshold. The situation where the consistency degree C is above the given threshold means that the winding device 200 has an abnormality or signs of an abnormality and requires maintenance work to be performed again.

[0187] In step S210, the notification determination unit 122 outputs the maintenance tasks to be notified to the notification unit 130. The maintenance tasks to be notified to the user are determined based on the equipment condition diagnosis model M, where the consistency C is above a given threshold.

[0188] In steps S211 and S212, the notification unit 130 notifies the user that maintenance work should be performed. In step S211, the alarm unit 131 issues an alarm. Furthermore, in step S212, the display unit 132 displays the details of the maintenance work notified to the user. Figure 10 Although an example is shown of both issuing an alarm in step S211 and displaying the content of the maintenance operation in step S212, it is also possible, for example, to not issue an alarm and only display the content of the maintenance operation.

[0189] The operator who receives the notification in steps S211 and S212 performs maintenance work on the winding device 200 based on the content of the notified maintenance work.

[0190] The processes described above, from steps S27 to S212, are the identification processes for maintaining the display device 100. Additionally, Figure 10 The identification process shown in steps S27 to S212 is related to... Figure 9 The identification processes shown in steps S17 to S112 are essentially the same.

[0191] <Details of each process>

[0192] The following is about Figure 9 as well as Figure 10 The learning process, recognition process, and update process are explained in detail.

[0193] [Learning Processing]

[0194] First, the learning process performed by the maintenance effect determination unit 123 and the equipment condition diagnosis model generation unit 124 will be explained.

[0195] (Processing of maintenance effect assessment department 123)

[0196] The following describes the processing performed by the maintenance effect judgment unit 123 in the learning process. Figure 9 The process of steps S11 to S14 will be explained. Figure 11 This is a flowchart used to explain the processing performed by the maintenance effect judgment unit 123 in the learning process.

[0197] In step S31, the maintenance effect determination unit 123 reads from the production actual results database 111 a production actual results data list PL that includes all production actual results data of the wound body 204 produced within a given time period from the time the maintenance operation was performed before the learning process, which is among the production actual results data registered in the production actual results database 111. before The given time is a pre-set length of time required to manufacture a certain number of windings 204.

[0198] In step S32, the maintenance effect determination unit 123 determines the maintenance effect based on the production actual result data list PL. before The actual production results data included are used to calculate the pre-maintenance defect rate Nf. before As mentioned above, the failure rate Nf before maintenance before PL is a list of actual production results data. before The shape data of the actual production results included, as well as the number of windings 204 that were judged to be defective by inspection results, are calculated by dividing the total number of productions before maintenance.

[0199] In step S33, the maintenance effect determination unit 123 reads from the production actual result database 111 a production actual result data list PL that includes all production actual result data of the wound body 204 produced from the time of maintenance to a given time thereafter. after .

[0200] In step S34, the maintenance effect determination unit 123 determines the maintenance effect based on the production actual result data list PL. after The actual production results data included are used to calculate the post-maintenance defect rate Nf. after As mentioned above, the post-maintenance failure rate Nf after PL is a list of actual production results data. after The shape data of the actual production results included, as well as the number of windings 204 that were judged to be defective by inspection results, are calculated by dividing the total production after maintenance operations.

[0201] In step S35, the maintenance effect determination unit 123 takes the defect rate Nf before maintenance. before Defect rate Nf after maintenance after The difference is used to determine whether the difference is greater than a given threshold Th. N When the difference is greater than the threshold Th N If the condition is met (step S35: yes), the maintenance effect determination unit 123 proceeds the process to step S36; otherwise (step S35: no), the process proceeds to step S37.

[0202] In step S36, due to the defect rate Nf before maintenance before Compared to the post-maintenance failure rate Nf after The size decreased, therefore the maintenance effect assessment unit 123 determined that the maintenance operation was effective. The maintenance operation mentioned here refers to the operation performed before the learning process, i.e., compared to... Figure 9 The maintenance work performed in step S11 is the first step.

[0203] On the other hand, in step S37, due to the defect rate Nf before maintenance... before Compared to the post-maintenance failure rate Nf after Since the size did not decrease, the maintenance effect assessment unit 123 determined that the maintenance operation was ineffective or had a very small effect.

[0204] In this way, the maintenance effect judgment unit 123 determines whether the maintenance work performed before the learning process is effective.

[0205] Figure 12A as well as Figure 12B This is a conceptual diagram used to illustrate the methods for judging the effectiveness of maintenance tasks in learning processes. Specifically, Figure 12A Examples are shown where the maintenance work is deemed effective. Figure 12B Examples are shown where maintenance operations are deemed ineffective. Figure 12A as well as Figure 12B The image shows five windings wound on one of a plurality of cores 206.

[0206] exist Figure 12A as well as Figure 12B In the example shown, before maintenance, out of the five windings wound onto a certain core 206, two were deemed defective. That is, the defect rate Nf before maintenance was... before It is 40%. Figure 12A In the example shown, after maintenance, among the five windings wound onto a certain core 206, the number of defective windings becomes 0 (the defect rate Nf after maintenance). after =0). On the other hand, in Figure 12B In the example shown, after maintenance, among the five windings wound onto a certain core 206, the number of windings judged as defective remained unchanged compared to before maintenance, becoming two (the defect rate Nf after maintenance). after =40%).

[0207] Therefore, in Figure 12A In the example shown, the defect rate Nf before maintenance before Defect rate Nf after maintenance after The difference is 40%. On the other hand, in Figure 12B In the example shown, the defect rate Nf before maintenance before Defect rate Nf after maintenance after The difference is 0. Therefore, for example, if the threshold ThN for determining the presence or absence of maintenance effect is, for example, 20%, then... Figure 12A In the example shown, the maintenance operation was determined to be effective. Figure 12B In the example shown, the maintenance operation was determined to be ineffective.

[0208] (Processing of the equipment condition diagnostic model generation unit 124)

[0209] The following describes the processing performed by the device status diagnosis model generation unit 124 in the learning process. Figure 9 The process of steps S15 and S16 will be explained. Figure 13 This is a flowchart illustrating the process performed by the device status diagnostic model generation unit 124 during the learning process.

[0210] In step S41, the equipment condition diagnosis model generation unit 124 reads the actual maintenance result data MD of the maintenance operation that was determined to be effective in the maintenance effect determination unit 123.

[0211] In step S42, the equipment condition diagnosis model generation unit 124 reads the production actual results data list PL before maintenance from the production actual results database 111. before Additionally, here is the list of actual production results data before maintenance, read by the equipment condition diagnostic model generation unit 124. before The list of actual production results data before maintenance, read from the maintenance effect assessment unit 123. before Same (refer to) Figure 11 Step S31).

[0212] In step S43, the equipment condition diagnostic model generation unit 124 uses the read pre-maintenance production actual results data list PL before The corresponding shape data group generates a swapped data group containing swapped datasets of multiple datasets within the shape data group, where the shape data of each dataset has been interchanged. Additionally, the so-called actual result data list PL... before The corresponding shape data group means that it corresponds to the actual result data list PL before The shape data group containing the shape data group ID corresponds to the shape data group. In the following description, this will be compared with the list of actual production results data read before maintenance (PL). before The corresponding shape data group is recorded as data group 1.

[0213] The data groups are generated as follows. Let the first data group contain a dataset DSα representing the correspondence between core 206α and image Iα, a dataset DSβ representing the correspondence between core 206β and image Iβ, and a dataset DSγ representing the correspondence between core 206γ and image Iγ. Furthermore, let image Iγ exhibit defects in the winding 204γ.

[0214] In this case, the swapped dataset is generated by swapping the correspondence between core and shape data in multiple datasets.

[0215] Specific examples will be provided to illustrate this. Figure 14 This is a diagram used to illustrate the swapping of data groups. In Figure 14 The image shows a pattern where five swapped data groups, from the first data group to the sixth data group, are generated based on the first data group. Figure 14 The first data group shown is... Figure 5 The shape data group shown at the bottom is the same. In the first data group, image Iα and image Iβ are "good", and image Iγ is "bad". Additionally, in... Figure 14 In order to distinguish between image Iα and image Iβ, image Iα is shown as "Good 1" and image Iβ is shown as "Good 2".

[0216] The second data group is a data group in which images Iα and Iγ in the datasets DSα and DSγ contained in the first data group are interchanged. That is, in the second data group, image Iα of dataset DSα is replaced with image Iγ, and image Iγ of dataset DSγ is replaced with image Iα.

[0217] Furthermore, in the following explanations, the datasets generated based on datasets DSx and DSy will be referred to as the swapped datasets, denoted as DSxy and DSyx. Additionally, x and y are any of α, β, and γ, and x ≠ y. The swapped dataset DSxy represents the correspondence between convolutional core 206x and image Iy, and the swapped dataset DSys represents the correspondence between convolutional core 206y and image Ix.

[0218] Thus, in the second data group, a swapped dataset DSαγ was created between image Iγ (representing "defective") and core 206α, and a swapped dataset DSγα was created between image Iα (representing "good") and core 206γ. In the second data group, dataset DSβ was not swapped.

[0219] Data group 3 consists of datasets DSβ and DSγ included in data group 1, in which images Iβ and Iγ are interchanged. Specifically, in data group 3, a swapped dataset DSβγ is created between image Iγ (representing "poor" quality) and core 206β, and a swapped dataset DSγβ is created between image Iβ (representing "good" quality) and core 206γ. Data group 3 does not swap dataset DSα.

[0220] Data group 4 consists of datasets DSα and DSβ included in data group 1, in which images Iα and Iβ are interchanged. Specifically, in data group 4, a swapped dataset DSαβ is created between image Iβ (representing "Good 2") and core image 206α, and a swapped dataset DSβα is created between image Iα (representing "Good 1") and core image 206β. In data group 3, dataset DSγ was not swapped.

[0221] Data group 5 is a dataset where images Iα and Iβ in datasets DSα and DSβ (included in data group 1) are swapped, and images Iα and Iγ in datasets DSαβ and DSγ are swapped. Specifically, data group 5 includes a swapped dataset DSαγ corresponding to image Iγ ("poor") and core 206α, a swapped dataset DSβα corresponding to image Iα ("good 1") and core 206β, and a swapped dataset DSγβ corresponding to image Iβ ("good 2") and core 206γ.

[0222] Data group 6 is a dataset that, after swapping images Iβ and Iγ in datasets DSβ and DSγ contained in data group 1, further swapped images Iγ and Iα in datasets DSγβ and DSα. Specifically, data group 6 includes a swapped dataset DSαβ corresponding to image Iβ ("Good 2") and core image 206α, a swapped dataset DSβγ corresponding to image Iγ ("Poor") and core image 206β, and a swapped dataset DSγα corresponding to image Iα ("Good 1") and core image 206γ.

[0223] Thus, swapped data groups are generated by swapping the images in the datasets contained in the original data group.

[0224] Thus, in step S43, the equipment condition diagnostic model generation unit 124 utilizes the read pre-maintenance production actual results data list PL before The corresponding first data group generates a swapped data group, which interchanges the combinations of multiple cores 206 and shape data (images). Thus, a shape data group is generated from all combinations of multiple cores 206 and shape data (images).

[0225] In step S44, the equipment condition diagnostic model generation unit 124 uses the read maintenance actual result data MD, the first data group, and the replacement data group to generate a new equipment condition diagnostic model M. new .

[0226] The resulting device condition diagnostic model M new Multiple learned models are formed, representing the causes of defects in the shape data group used in generation, with the core 206 corresponding to the "defective" image being the cause of the defect.

[0227] Specific examples will be provided to illustrate this. For instance... Figure 14 As shown, the equipment condition diagnostic model M is generated using the first data group that was not swapped (image Iγ is "defective"). new This is a learned model representing the cause of the defect in the core 206γ. Therefore, the learned model generated below, using the shape data (image Iγ) of the winding 204γ that actually caused the defect and the core 206γ that is the cause of the defect, and the corresponding dataset DSγ, is recorded as the first equipment condition diagnostic model.

[0228] Moreover, the following will be as follows: Figure 14The model generated by creating the corresponding swap dataset DSαγ or DSβγ using the winding core 206α or 206β with actual non-defective windings 204α or 204β and the shape data (image Iγ) representing defects, as shown in the second to sixth data groups, is recorded as the second device condition diagnosis model.

[0229] use Figure 14 The device condition diagnostic model M was generated from the first data group shown (image Iγ and core 206γ established a corresponding dataset DSγ). new This is the first learned model indicating the cause of the defect in core 206γ. Similarly, using... Figure 14 The equipment condition diagnostic model M was generated from the fourth data group shown (image Iγ and core 206γ established a corresponding dataset DSγ). new This is the first learned model indicating the cause of the defect in core 206γ.

[0230] On the other hand, utilizing Figure 14 The device condition diagnostic model M is generated from the second data set shown (which includes image Iγ and core 206α to establish a corresponding dataset DSαγ). new This is the second learned model indicating the reason why core 206α is defective. (Utilizing...) Figure 14 The device condition diagnostic model M is generated from the third data group shown (which includes the corresponding dataset DSβγ established by image Iγ and core 206β). new This is the second learned model indicating the cause of the defect in core 206β. Utilizing... Figure 14 The equipment condition diagnostic model M generated from the fifth data group shown (image Iγ and core 206α are correlated) new This is the second learned model indicating the reason why core 206α is defective. (Utilizing...) Figure 14 The equipment condition diagnostic model M generated from the 6th data group shown (image Iγ and core 206β are correlated) new This is the second learned model indicating the reason why core 206β is defective.

[0231] Thus, in step S44, the equipment condition diagnosis model generation unit 124 uses a first data group containing a dataset DSγ representing a defect in the actual winding body 204γ to generate a first equipment condition diagnosis model indicating that the cause of the defect is the core 206γ. Simultaneously, the equipment condition diagnosis model generation unit 124 also uses swap data groups (data groups 2 to 6) containing swap datasets to generate a second equipment condition diagnosis model assuming a defect in the winding body 204α or 204β, indicating that the cause of the defect is the core 206α or 206β. Through this process, more learned models can be generated using less teaching data (represented as images of defects).

[0232] In step S45, the equipment condition diagnostic model generation unit 124 generates all the newly generated equipment condition diagnostic models M. new Registered to the equipment status diagnostic model database 112.

[0233] In this way, during the learning process, a new equipment condition diagnostic model M is generated, which learns what shape data defective products are improved through what maintenance operations. new And register it in the equipment status diagnostic model database 112.

[0234] Furthermore, during the learning process, when a shape data group indicating that a defect has actually occurred in the winding body wound on any one of the multiple cores 206 is obtained, a swap data group containing swap datasets is generated, assuming that the defect has occurred in the winding body wound on other cores. Then, using the shape data group indicating that a defect has actually occurred in the winding body wound on any one of the multiple cores 206, a device condition diagnostic model M is generated, indicating that the defect is caused by that core. new Furthermore, it utilizes a data exchange group to generate an equipment condition diagnostic model M, which represents the cause of the defect as the core corresponding to the other windings, assuming that other windings have produced defects. new .

[0235] The following describes a specific example of the learning process in the case of obtaining a first data group containing an image Iγ indicating that the wound body 204γ has produced a defect. Here, the defective wound body 204γ is an example of the second wound body of this disclosure, and the core 206γ is an example of the second core of this disclosure. The shape data used to generate the image Iγ is an example of the third and fourth sets of data of this disclosure. On the other hand, the wound body 204α or 204β that has not produced a defect is an example of the first wound body of this disclosure, and the core 206α or β is an example of the first core of this disclosure. The first piece position data group in the image Iα or Iβ (refer to...) Figure 4CThe first set of data in this disclosure is an example. The second set of positional data in image Iα or Iβ is an example of the second set of data.

[0236] The equipment condition diagnosis model generation unit 124 generates a first equipment condition diagnosis model indicating the cause of the defect of the core 206γ by using the first data group or the fourth data group of the dataset DSγ containing the correspondence between the core 206γ and the image Iγ through the above learning process.

[0237] Furthermore, the equipment condition diagnosis model generation unit 124, through the aforementioned learning process, utilizes swapped data groups (second, third, fifth, and sixth data groups) containing swapped datasets in which images of the datasets included in the first data group have been swapped, and assumes that a new shape data group containing shape data showing a defect in the wound body 204α or 204β has been acquired, generates a second equipment condition diagnosis model for determining the cause of the defect in the core 206α or 206β. In this case, the newly acquired defective wound body 204α or 204β is assumed to be an example of the third wound body of this disclosure. Moreover, the first piece position data group (refer to) in the image Iα or Iβ of the newly acquired defective wound body 204α or 204β is assumed to be... Figure 4C This is an example of the fifth set of data in this disclosure. The second position data set in the newly acquired image 1α or 1β containing the defective winding 204α or 204β is assumed to be an example of the sixth set of data in this disclosure.

[0238] The datasets DSα and DSβ, which consist of convolution core 206α and image Iα respectively, and DSβ, which consist of convolution core 206β and image Iβ respectively, are examples of the first dataset of this disclosure. The dataset DSγ, which consists of convolution core 206γ and image Iγ respectively, is an example of the second dataset of this disclosure.

[0239] Furthermore, in the swapped data group containing images Iα and Iγ swapped between datasets DSα and DSγ, the swapped dataset DSαγ established by core 206α and image Iγ is an example of the first swapped dataset of this disclosure. Similarly, in the swapped data group containing images Iβ and Iγ swapped between datasets DSβ and DSγ, the swapped dataset DSβγ established by core 206β and image Iγ is an example of the first swapped dataset of this disclosure. Moreover, in the swapped data group containing images Iα and Iγ swapped between datasets DSα and DSγ, the swapped dataset DSγα established by core 206γ and image Iα is an example of the second swapped dataset of this disclosure. Similarly, in the swapped data group containing images Iβ and Iγ swapped between datasets DSβ and DSγ, the swapped dataset DSγβ established by core 206γ and image Iβ is an example of the second swapped dataset of this disclosure.

[0240] Furthermore, in the learning process described above, the equipment status diagnostic model generation unit 124 generates a first equipment status diagnostic model using the first and fourth data groups containing the second dataset, and generates a second equipment status diagnostic model using the second, third, fifth, and sixth data groups containing the swapped dataset.

[0241] Through the learning process of this disclosure, the following effect can be achieved. In cases where the proportion of defective products generated in the winding device 200 is low, it can be difficult to aggregate the teaching data that can accurately determine the cause of the defect. Even in such cases, according to the learning process of this disclosure, multiple replacement data groups can be generated using shape data groups containing shape data representing the defect, and these can be used as teaching data to generate the equipment condition diagnostic model M. Therefore, it is possible to generate an equipment condition diagnostic model M where all multiple windings 206 are defective using only the shape data group where any one of the multiple windings 206 is the cause of the defect.

[0242] According to this method for generating the equipment condition diagnostic model M, the teaching data used to generate the equipment condition diagnostic model M can be increased substantially, thus enabling the generation of a larger number of equipment condition diagnostic models M. Furthermore, the following beneficial effect can be obtained: the diagnostic accuracy of subsequent identification processing utilizing the causes of defects identified by the equipment condition diagnostic model M is easily improved. Moreover, as... Figure 14 As shown, in the case of multiple "good" shape data, by distinguishing the "good" from each other, the group of shape data that can be used as teaching data can be further increased compared with the case where no distinction is made.

[0243] [Identification Processing]

[0244] The identification process performed by the equipment status diagnosis unit 121 and the notification determination unit 122 will be described below.

[0245] (Processing of Equipment Status Diagnosis Unit 121)

[0246] The following describes the processing performed by the device status diagnostic unit 121 during the identification process. Figure 9 The process of steps S17 and S18 will be explained. Figure 15 This is a flowchart explaining the process performed by the device status diagnostic unit 121 during the identification process.

[0247] In step S51, the equipment status diagnostic unit 121 determines whether new production actual result data PD has been registered in the production actual result database 111. new No new production data (PD) has been registered. new In the case of (step S51: No), the equipment status diagnostic unit 121 repeatedly executes step S51. After registering new actual production result data PD... new If the condition is met (step S51: Yes), the device status diagnostic unit 121 proceeds the process to step S52.

[0248] In step S52, the equipment status diagnostic unit 121 performs diagnostics based on the newly registered actual production result data PD. new Extract the production actual results data list PL from the production actual results database 111. The production actual results data list PL is the newly registered production actual results data PD from the production actual results data PD registered in the production actual results database 111. new The production actual results data (PD) of the wound bodies 204 produced within a given time period from the production date is extracted and listed. That is, the production actual results data list (PL) contains at least the newly registered production actual results data (PD). new .

[0249] In step S53, the equipment condition diagnosis unit 121 generates a consistency degree C using the shape data contained in the newly registered production actual result data list PL and the past shape data contained in the equipment condition diagnosis model M read from the equipment condition diagnosis model database 112. The equipment condition diagnosis model used here is any one of a plurality of equipment condition diagnosis models that includes the first equipment condition diagnosis model and the second equipment condition diagnosis model, generated through the learning process described above.

[0250] More specifically, the equipment condition diagnostic unit 121 extracts shape data from one or more actual production result data contained in the actual production result data list PL (see reference). Figure 5On the other hand, the equipment condition diagnosis unit 121 extracts multiple equipment condition diagnosis models M registered in the equipment condition diagnosis model database 112. Each of the multiple equipment condition diagnosis models M corresponds to a different cause of failure, and further, to a different maintenance operation.

[0251] The equipment condition diagnosis unit 121 calculates multiple consistency degrees C from all combinations of shape data extracted from one or more actual production results data and multiple equipment condition diagnosis models M.

[0252] (Notify the Judgment Department 122 of the processing)

[0253] The following describes the processing performed by the notification determination unit 122 in the identification process. Figure 9 The process of steps S19 to S112 will be explained. Figure 16 This is a flowchart for explaining the process performed by the determination unit 122 in the identification process.

[0254] In step S61, the notification determination unit 122 sums up the consistency scores C for each maintenance group based on the multiple consistency scores C generated by the equipment status diagnosis unit 121. A maintenance group is a group corresponding to the content of the maintenance work. For example, the notification determination unit 122 sums up the consistency scores C for each maintenance group based on the multiple consistency scores C generated by the equipment status diagnosis unit 121. Figure 17A The maintenance team information shown organizes the maintenance tasks performed into smaller teams. For example... Figure 17A As shown, the maintenance team information establishes a corresponding relationship between the maintenance team and the maintenance tasks performed within that team. For example... Figure 17A As shown, the maintenance team information may further include the maintenance plan to be implemented corresponding to the maintenance team. Furthermore, in this embodiment, a maintenance team is described as a group divided according to each component of the maintenance operation object; however, the invention is not limited to this, and the maintenance team may also be divided according to each content of the maintenance operation, or according to each model of the parts replaced in the maintenance operation, etc.

[0255] In the following explanation, the total value A will be defined as the sum of the consistency C for each maintenance team. The method for generating the total value A can be appropriately chosen from several types of summation methods. Specific examples of these summation methods include, for instance, summing only the consistency C, averaging the consistency C, selecting the maximum value from the consistency C, and averaging a given number of consistency C values ​​from the higher level.

[0256] In step S62, the determination unit 122 is notified to generate a maintenance plan list ML. The maintenance plan list ML is a list of maintenance teams, for example, the maintenance teams are arranged in descending order of their total value A. Figure 17BThis is a diagram showing a specific example of the maintenance scheme list ML.

[0257] like Figure 17B As shown, the maintenance plan list ML contains data such as "Maintenance Plan ID," "Equipment," "Maintenance Plan," and "Total Value." The "Maintenance Plan ID" data is an identifier assigned to each maintenance team, rearranged according to the total value. For example, the larger the total value, the smaller the number assigned to the "Maintenance Plan ID." The "Maintenance Plan" data represents the maintenance content to be performed for the corresponding maintenance team. The notification determination unit 122 refers to this. Figure 17A The maintenance team information shown is used to determine the maintenance plan corresponding to each maintenance team. The "Total Value" data represents the total value A for each maintenance team.

[0258] exist Figure 17B In the example shown, in the winding device "A", the maintenance of the first winding of one of the multiple winding cores 206 is registered as maintenance group 1, and the maintenance of the third winding of the other multiple winding cores 206 is registered as maintenance group 2. Moreover, the maintenance of the second winding of the other multiple winding cores 206 is registered in the maintenance plan list ML as maintenance group 3.

[0259] In addition, the so-called Figure 17A The "maintenance of the first core" in the maintenance plan shown in Figure B means that it includes at least one of the maintenance operations on the first core, such as adjusting the first core, cleaning the first core, or replacing the first core. The same applies to "maintenance of the third core" and "maintenance of the second core".

[0260] For reference Figure 5 As explained, if the continuity of the position of the upper end face of the first sheet 202 indicated by the first sheet end position data group and the continuity of the position of the upper end face of the second sheet 203 indicated by the second sheet end position data group become inclined from the baseline, the winding body 204 is judged to be defective (or acceptable). Moreover, it is known that the reason for the defect of the winding body 204 is the core 206 on which the winding body 204 is wound.

[0261] The total value A is the sum of the consistency levels C, and therefore has the same properties as the consistency level C. Therefore, the larger the total value A, the greater the necessity for the maintenance work performed by the maintenance team for the winding device 200. Furthermore, the maintenance plan list ML is a list of maintenance teams arranged in descending order of the total value A; therefore, the higher the maintenance team is in the maintenance plan list ML, the greater the necessity for maintaining the winding device 200.

[0262] In step S63, the determination unit 122 is notified to determine whether the total value A for each maintenance team is greater than the given warning threshold Th. f The so-called given warning threshold Th f This is the minimum value of the total number of signs that an anomaly has occurred in the winding device 200. Furthermore, in this embodiment, an anomaly in the winding device 200 means, for example, that the winding device 200 produces a given percentage or more of the wound bodies 204 with an inspection result of "defective". Conversely, a sign that an anomaly in the winding device 200 is occurring means, for example, that the winding device 200 produces a given percentage or more of the wound bodies 204 with an inspection result of "pass". Therefore, if the total value A is less than the sign threshold TH... f Therefore, it can be expected that the proportion of "good" inspection results for subsequently produced wound bodies 204 will be above a given proportion. The given warning threshold Th f For example, decisions can be made based on experience, such as past maintenance data (MD).

[0263] Even if the maintenance plan list ML contains only one total value A greater than the warning threshold Th f In the case of the maintenance team (step S63: Yes), the notification determination unit 122 also advances the processing to step S64. When the total value A is greater than the warning threshold Th... f If none of the maintenance teams are included in the maintenance plan list ML (step S63: No), the notification determination unit 122 considers it as not requiring maintenance and ends the process.

[0264] In step S64, the determination unit 122 is notified to determine whether there is a total value A among the maintenance groups included in the maintenance plan list ML that is greater than a given abnormal threshold Th. a The maintenance team. The so-called given anomaly threshold Th a Th refers to the minimum total value of any abnormality that would occur in the winding device 200 if the warning stage had been exceeded. Therefore, the abnormality threshold Th... a For example, based on past maintenance data such as MD, it is determined through experience that the threshold Th is greater than the warning threshold. f The value. When the total value A is greater than the abnormal threshold Th. a If the maintenance team is included in the maintenance plan list ML (step S64: Yes), the determination unit 122 is notified to proceed to step S66. When the total value A is greater than the abnormal threshold Th... a If the maintenance team is not included in the maintenance plan list ML (step S64: No), notify the decision unit 122 to proceed to step S65.

[0265] In step S65, the notification determination unit 122 causes the display unit 132 of the notification unit 130 to notify that the total value A is determined to be greater than the warning threshold Th in step S63. f The maintenance tasks corresponding to the maintenance team. More specifically, the notification determination unit 122, for example, causes the display unit 132 to display not only a message such as "Please perform the following maintenance tasks," but also the recommended maintenance tasks. Furthermore, the recommended maintenance tasks are... Figure 17B The content corresponding to the "Maintenance Plan" data contained in the maintenance plan list ML shown.

[0266] Here, the notification determination unit 122 notifies that there are multiple total values ​​A greater than the warning threshold Th. f In the case of a maintenance team, the contents of multiple maintenance tasks can also be sorted and displayed according to the total value. In this case, more detailed notification to the determination unit 122 is provided, such as "Please perform the following maintenance tasks. If performing the superior maintenance tasks does not improve the situation, sometimes performing the subordinate maintenance tasks can improve the situation." and multiple recommended maintenance tasks can be displayed in order from superior to subordinate.

[0267] Furthermore, the notification determination unit 122 not only notifies the maintenance work content but also notifies the maintenance team that has established a corresponding maintenance plan ID. When the operator who performed the maintenance work inputs the actual maintenance result data MD, by establishing a correspondence between the actual maintenance result data MD and the maintenance plan ID that triggered the maintenance, it is easy to determine whether the input actual maintenance result data MD corresponds to the maintenance work performed based on the notification from the maintenance display device 100.

[0268] In step S66, the notification determination unit 122, similar to step S65, causes the display unit 132 to display the maintenance operation content, and the alarm unit 131 to issue an alarm notifying the user of the maintenance display device 100 of an abnormality. When the winding device 200 of the object experiences an abnormality rather than any signs of an abnormality, it is a situation requiring urgent maintenance. Therefore, the notification determination unit 122 not only displays the maintenance operation content based on the display unit 132, but also issues an alarm through the alarm unit 131, quickly notifying the user of the maintenance display device 100 of the occurrence of the abnormality.

[0269] In this way, during the identification process, the actual production result data PD of the newly produced winding 204 (especially the shape data in the shape data group indicating "defective" or "passable") and the equipment condition diagnostic model M are used to determine whether an anomaly (a situation where a given proportion of defective products are produced) or a sign of an anomaly has occurred in any of the cores 206. Furthermore, if an anomaly or a sign of an anomaly is determined to have occurred, the user is notified. Thus, the user can quickly learn of any anomaly occurring in any of the cores 206 and can learn about the maintenance work that should be performed to mitigate the anomaly.

[0270] Furthermore, as described above, in this disclosure, it is assumed that the defect of the winding body 204 is caused by the core 206 on which the winding body 204 is wound. In the maintenance display device 100, as described above, the total value of each maintenance group is calculated, with any one of the multiple cores 206 as the object of maintenance work, and the maintenance work of which core 206 should be performed is determined based on the magnitude of the total value. Through such determination, among any one of the multiple cores 206, the maintenance work with a high probability of eliminating the defect of the winding body 204 by performing the maintenance work is extracted and displayed.

[0271] [Update Processing]

[0272] The update process performed by the maintenance effect determination unit 123 and the equipment status diagnosis model generation unit 124 will be described below.

[0273] (Processing of maintenance effect assessment department 123)

[0274] The following describes the processing performed by the maintenance effect determination unit 123 during the update process. Figure 10 The process of steps S21 to S24 will be explained. Figure 18 This is a flowchart used to explain the processing performed by the maintenance effect determination unit 123 in the update process.

[0275] In step S71, the maintenance effect determination unit 123 determines whether new maintenance actual result data MD has been registered in the maintenance actual result database 113 of the storage unit 110. new The data was determined to be unregistered, and no new maintenance results data (MD) was recorded. new If (step S71: No), the maintenance effect determination unit 123 repeatedly executes step S71. If it is determined that new maintenance actual result data MD has been registered... new If the condition is met (step S71: yes), the maintenance effect determination unit 123 advances the processing to step S72.

[0276] In step S72, the maintenance effect determination unit 123 determines the maintenance effect based on the newly registered maintenance actual result data MD.new Includes "maintenance date and time" data (refer to) Figure 8 The determination is based on the actual maintenance results data (MD) of the newly registered data. new Whether the corresponding maintenance has been performed within the given time frame.

[0277] If it is determined that a given time has elapsed since the start of the maintenance operation (step S72: Yes), the maintenance effect determination unit 123 proceeds to step S73. If it is determined that no given time has elapsed since the start of the maintenance operation (step S72: No), the maintenance effect determination unit 123 repeatedly executes the process of step S72.

[0278] In step S73, the maintenance effect determination unit 123 reads from the production actual results database 111 a list of pre-maintenance production actual results data PL, which includes all production actual results data PD of the wound body 204 produced from the maintenance operation to a previously given time. before .

[0279] In step S74, the maintenance effect determination unit 123 reads the new maintenance actual result data MD from the equipment condition diagnosis model database 112. new The corresponding maintenance team's equipment status diagnostic model M, and based on the read equipment status diagnostic model M and the actual production result data list PL. before To generate consistency C before maintenance before Regarding consistency C before maintenance before The generation method is the same as... Figure 15 The method for generating consistency C, which is performed by the device status diagnostic unit 121 in step S53, is the same.

[0280] In step S75, the maintenance effect determination unit 123 reads from the production actual result database 111 a production actual result data list PL that includes all production actual result data PD of the wound body 204 produced from the maintenance operation to a given time thereafter. after .

[0281] In step S76, the maintenance effect determination unit 123 reads the new maintenance actual result data MD from the equipment condition diagnosis model database 112. new The corresponding maintenance team's equipment status diagnostic model M, and based on the read equipment status diagnostic model M and the actual production result data list PL. after To generate the maintained consistency C after Regarding consistency C after The generation method is the same as... Figure 15The method for generating consistency C, which is performed by the device status diagnostic unit 121 in step S53, is the same.

[0282] In step S77, the maintenance effect determination unit 123 takes the consistency C before maintenance. before Consistency C with the maintained after The difference is used to determine whether the difference is greater than a given value Th. D When the difference is greater than the given value Th D If the condition is met (step S77: Yes), the maintenance effect determination unit 123 proceeds the process to step S78; otherwise (step S77: No), the process proceeds to step S79. Given a threshold Th D The decision can be made appropriately based on the actual results of past maintenance work.

[0283] In step S78, due to the consistency C with before maintenance before Compared to the consistency C after maintenance after The size decreases, therefore the maintenance effect determination unit 123 determines that the maintenance work performed based on the maintenance content notified by the notification determination unit 122 is effective.

[0284] In step S79, due to the consistency C with before maintenance... before Compared to the consistency C after maintenance after If the value decreases, the maintenance effect determination unit 123 determines that the maintenance operation performed based on the maintenance content notified by the notification determination unit 122 is ineffective or has a very small effect.

[0285] Figure 19A as well as Figure 19B This is a conceptual diagram used to illustrate the methods for judging the effectiveness of maintenance operations during the update process. Specifically, Figure 19A Examples are shown where the maintenance work is deemed effective. Figure 19B Examples are shown where maintenance is deemed ineffective.

[0286] exist Figure 19A as well as Figure 19B In the example shown, the pre-maintenance consistency C is calculated using the shape data set of the pre-maintenance wound body 204 and the equipment condition diagnostic model M. before =0.90.

[0287] Moreover, in Figure 19A In the example shown, the post-maintenance consistency C is calculated using the shape data set of the wound body 204 produced after maintenance and the equipment condition diagnostic model M. after =0.20. On the other hand, in Figure 19BIn the example shown, the post-maintenance consistency C is calculated using the shape data set of the wound body 204 produced after maintenance and the equipment condition diagnostic model M. after =0.90.

[0288] Therefore, in Figure 19A In the example shown, the consistency degree C before maintenance is... before Consistency C after maintenance after The difference becomes 0.70. On the other hand, in Figure 19B In the example shown, the consistency degree C before maintenance is... before Consistency C after maintenance after The difference becomes 0. Therefore, for example, in determining the presence or absence of maintenance effect threshold Th D When the value is 0.30, in Figure 19A In the example shown, the maintenance operation was determined to be effective. Figure 19B In the example shown, the maintenance operation was determined to be ineffective.

[0289] (Processing of the equipment condition diagnostic model generation unit 124)

[0290] The following describes the processing performed by the device status diagnostic model generation unit 124 during the update process. Figure 10 The process of steps S25 and S26 will be explained. Figure 20 This is a flowchart illustrating the process performed by the device status diagnostic model generation unit 124 during the update process.

[0291] In step S81, the equipment condition diagnostic model generation unit 124 reads the actual maintenance result data MD of the maintenance operation that was determined to be effective in the maintenance effect determination unit 123. new .

[0292] In step S82, the equipment condition diagnosis model generation unit 124 reads the production actual results data list PL before maintenance from the production actual results database 111. before Additionally, here is the list of actual production results data before maintenance, read by the equipment condition diagnostic model generation unit 124. before The list of actual production results data before maintenance, read from the maintenance effect assessment unit 123. before Same (refer to) Figure 11 Step S31).

[0293] In step S83, the equipment condition diagnostic model generation unit 124 uses the read pre-maintenance production actual results data list PL before For the corresponding first data group, generate a swapped data group by interchanging the image data of the multiple datasets contained in the first data group. Step S83 is related to... Figure 13 The process shown in step S43 is the same.

[0294] In step S84, the equipment condition diagnostic model generation unit 124 uses the read maintenance actual result data MD, the first data group, and any one of the swapped data group to generate a new equipment condition diagnostic model M. new .

[0295] In step S85, the equipment condition diagnostic model generation unit 124 generates a new equipment condition diagnostic model M. new Update the equipment status diagnostic model M that has already been registered in the equipment status diagnostic model database 112.

[0296] Thus, in the update process, a new equipment condition diagnostic model M is generated using the equipment condition diagnostic model M generated in the learning process. new And utilize the new equipment condition diagnostic model M new This updates the equipment condition diagnostic model M that has been registered in the equipment condition diagnostic model database 112. In this way, the new equipment condition diagnostic model M is utilized based on effective maintenance operations. new The equipment status diagnosis model M in the equipment status diagnosis model database 112 is updated, thereby gradually improving the diagnostic accuracy of the equipment status of the winding device 200 in the equipment status diagnosis unit 121.

[0297] The method for displaying information for maintenance of a production apparatus disclosed herein involves inputting, in a learned model created by the learned model generation method disclosed herein, a fifth set of data representing the position of the fifth end face of the first electrode sheet along the radial direction of a third winding body in which the first and second electrode sheets are wound in multiple overlapping turns on the first core, and a sixth set of data representing the position of the sixth end face of the second electrode sheet along the radial direction of the third winding body. Then, if, based on the continuity of the position of the fifth end face represented by the fifth set of data, the continuity of the position of the sixth end face represented by the sixth set of data, and the positional relationship of the reference line, it is determined that the third winding body is defective, and information indicating that the defect is caused by the first core is output from the learned model, the information indicating that the third winding body is defective and the defect is caused by the first core is output to a display device.

[0298] <Function and Effect of the Maintenance Display Device 100 in the First Embodiment>

[0299] As explained above, the maintenance display device 100 includes a notification determination unit 122 and a device status diagnostic model generation unit 124, which is an example of a model generation unit. The notification determination unit 122 acquires multiple sets of data indicating the position of the end face from an inspection machine 207, which is a sensor, read along the radial direction for each of the multiple windings 204 respectively wound on multiple cores 206. Then, the notification determination unit 122 determines whether the winding 204 is defective based on whether the continuity of the position of the first end face indicated by the first set of data intersects with the continuity of the position of the second end face indicated by the second set of data. If the winding 204 is defective, the notification determination unit 122 outputs information indicating that the defect is caused by any one of the multiple cores 206 to the display unit 132 for maintenance. Furthermore, the equipment condition diagnosis model generation unit 124 uses multiple sets of data to generate a shape dataset that establishes a corresponding relationship between any one of the multiple cores 206 and any one of the multiple sets of data, and generates a swapped dataset that swaps the correspondence between the cores 206 and the sets of data in all combinations of the multiple cores 206 and the multiple sets of data. Then, the equipment condition diagnosis model generation unit 124 uses this swapped dataset to generate or update multiple learned models M representing the cause of defects in any one of the multiple cores 206.

[0300] Thus, according to the maintenance display device 100 of the first embodiment, when a shape data group containing shape data (image Iy) of a wound body 204y that actually has defects and is wound in multiple cores 206x and 206y is acquired, a corresponding swap dataset is established between the shape data (image Iy) of the wound body 204y that actually has defects and the core 206x that is wound in a wound body 204x that does not actually have defects. Furthermore, a corresponding dataset DSy was established using the shape data (image Ix) of the winding 206y that actually produced a defective winding 204y among multiple windings 206, and the winding 204y that actually produced a defective winding. A device status diagnostic model M (first device status diagnostic model) indicating that the defect is caused by the winding 206y was generated. In addition, a corresponding swap dataset DSxy was established using the shape data (image Iy) of the winding 206x that actually did not produce a defective winding 204x and the winding 204y that actually produced a defective winding. A device status diagnostic model M (second device status diagnostic model) indicating that the defect is caused by the winding 206x was generated under the assumption that the shape data indicating that the winding 204x that did not actually produce a defect was obtained.

[0301] With this structure, it is possible to generate a device condition diagnostic model M in the case where all of the multiple cores 206 are defective, using only the shape data group where any one of the multiple cores 206 is the cause of the defect.

[0302] According to this method for generating the equipment condition diagnostic model M, the teaching data used to generate the equipment condition diagnostic model M can be increased substantially, thus enabling the generation of a larger number of equipment condition diagnostic models M. Furthermore, the following beneficial effect can be achieved: the diagnostic accuracy of subsequent identification processing utilizing the causes of defects identified by the equipment condition diagnostic model M is easily improved.

[0303] Furthermore, in the identification process, the maintenance display device 100 uses the equipment status diagnostic model M generated in this way to calculate the consistency C between the shape data of the winding body 204 produced after the maintenance operation and the equipment status diagnostic model M for each maintenance team, and determines whether to issue an alarm and notify the content of the maintenance operation or only notify the content of the maintenance operation based on the magnitude of the consistency C.

[0304] Furthermore, based on this structure, the status of the winding device 200 can be appropriately diagnosed using a learned model (equipment status diagnostic model M) generated from effective (defect rate reduction) maintenance operations performed in actual maintenance work. Moreover, since the learned model is updated continuously, the accuracy of the diagnosis can be improved. Furthermore, in the event of a diagnosed malfunction in the winding device 200, an alarm can be issued to prompt the user to take emergency action, and in the event of a diagnosed sign of an impending malfunction, the user can be notified of foreseeable improvements in maintenance operations. Therefore, maintenance operations can be performed when the defect rate in the winding device 200 is low.

[0305] In the maintenance display device 100 according to the first embodiment, any one of the multiple winding cores 206 is assumed as the cause of the defect generated in the winding body 204. In the maintenance display device 100 according to the first embodiment, the notification determination unit 122 sums the consistency C for each maintenance group and determines the content of the maintenance operation to be notified to the user based on the magnitude of the total value A. Therefore, the user is notified of the maintenance operation among the multiple winding cores 206 that has the highest probability of defect improvement through maintenance. In the case of multiple highly probable maintenance operations, the multiple maintenance operations are displayed in a sorted state. As a result, the user can appropriately improve the defect of the winding body 204 by performing the notified maintenance operations in descending order of rank.

[0306] The maintenance display device according to this embodiment includes a notification unit, a maintenance effect determination unit, and an equipment status diagnostic model generation unit. The notification unit notifies the user of the maintenance operation content based on an equipment status diagnostic model that is linked to and registered in a database for each past maintenance operation, along with previously entered production performance data, and newly input production performance data. The maintenance effect determination unit determines whether the maintenance operation was effective based on production performance data earlier and later than the time the maintenance operation was performed. The equipment status diagnostic model generation unit generates a new equipment status diagnostic model based on production performance data earlier than the time the effective maintenance operation was performed, and the content of the effective maintenance operation.

[0307] The maintenance display device according to this embodiment also includes: an equipment status diagnosis unit that generates equipment status diagnosis indicators, which are the degree of consistency between newly registered actual production results data and actual production results data before maintenance operations included in the equipment status diagnosis model. Furthermore, a notification unit notifies the user of maintenance operations based on the equipment status diagnosis indicators.

[0308] In the maintenance display device according to this embodiment, the equipment condition diagnosis model generation unit uses actual production results data earlier than the time point when the maintenance operation was determined to be effective, and actual maintenance results data related to the maintenance operation, to generate an equipment condition diagnosis model through machine learning.

[0309] The maintenance display device according to this embodiment performs maintenance operations not based on maintenance operations not notified by the notification unit. When new maintenance performance data related to this maintenance operation is input, it calculates the defect rate of defective products in the production performance data for the period from the time of the maintenance operation with the newly input maintenance performance data to a previously given time, based on data related to the inspection results of the products produced by the production equipment included in the production performance data. Furthermore, it calculates the defect rate of defective products in the production performance data for the period from the time of the maintenance operation with the newly input maintenance performance data to a subsequently given time. Then, the maintenance effect determination unit calculates the difference between the defect rate before the maintenance operation and the defect rate after the maintenance operation, and determines whether the maintenance operation was effective based on the magnitude of the difference.

[0310] (Second Implementation)

[0311] The second embodiment of this disclosure will now be described. Figure 21This is a diagram illustrating the structure of the maintenance display device 100A according to the second embodiment. In the maintenance display device 100A according to the second embodiment, the processing performed by the maintenance effect determination unit 123A of the control unit 120A of the server 10A is different from that of the maintenance effect determination unit 123 according to the first embodiment described above.

[0312] The differences from the first embodiment will be described below. Structures identical to those in the first embodiment will be indicated by the same symbols as those in the first embodiment, and structures different from those in the first embodiment will be indicated by the symbol "A".

[0313] In the first embodiment, it is not assumed that the user of the maintenance display device 100 will perform maintenance work other than that notified by the maintenance display device 100. However, in practice, in the use of the winding device 200, appropriate and necessary maintenance work (maintenance work other than that notified by the maintenance display device 100) may be performed at any time based on on-site judgment, etc. In this second embodiment, the maintenance display device 100A will also be described as being able to handle situations where maintenance work other than that notified by the maintenance display device 100A is performed.

[0314] Figure 22 This is a flowchart explaining the processing performed by the maintenance effect determination unit 123A in the second embodiment.

[0315] exist Figure 22 In step S91, the maintenance effect determination unit 123A determines whether maintenance actual result data MD has been newly registered in the maintenance actual result database 113 of the storage unit 110. new The data was determined to be unregistered, and no new maintenance results data (MD) was recorded. new If the condition is not met (step S91: No), the maintenance effect determination unit 123A repeatedly executes step S91. If it is determined that new maintenance actual result data MD has been registered... new If the condition is met (step S91: Yes), the maintenance effect determination unit 123A advances the process to step S92.

[0316] In step S92, the maintenance effect determination unit 123A determines the maintenance effect based on the newly registered maintenance actual result data MD. new The maintenance date and time data included are used to determine the actual maintenance results data (MD) from the newly registered data. new Whether a given time has elapsed since the corresponding maintenance operation began. The given time is the same as the given time described in the first embodiment, for example, the time required after the maintenance operation to manufacture a certain number of windings 204 in the object winding device 200.

[0317] If it is determined that a given time has elapsed since the start of the maintenance operation (step S92: Yes), the maintenance effect determination unit 123A proceeds the process to step S93. If it is determined that no given time has elapsed since the start of the maintenance operation (step S92: No), the maintenance effect determination unit 123A repeatedly executes the process of step S92.

[0318] In step S93, the maintenance effect determination unit 123A determines whether there is any data MD that matches the newly registered actual maintenance results. new A corresponding maintenance plan ID is established. As described in the first embodiment, the notification determination unit 122 notifies not only the maintenance work content but also the maintenance team associated with that maintenance content that a corresponding maintenance plan ID has been established. The operator performs the maintenance work represented by the notified maintenance plan ID. The operator establishes a correspondence between the performed maintenance work and the notified maintenance plan ID and inputs the maintenance actual result data MD. Thus, a correspondence is established between the maintenance actual result data MD and the maintenance plan ID that became the opportunity for maintenance. In this step S93, the newly registered maintenance actual result data MD is determined in this way. new Was the maintenance initiated based on a notification from the maintenance display device 100A?

[0319] In step S93, there exists a data MD that matches the newly registered maintenance actual results. new If a corresponding maintenance plan ID has been established, it is determined to be related to the actual maintenance result data MD. new The corresponding maintenance work is initiated by notifications based on the maintenance content displayed on the maintenance display device 100A. Furthermore, there is no data MD related to newly registered actual maintenance results. new If a corresponding maintenance plan ID has been established, it is determined to be related to the actual maintenance result data MD. new The corresponding maintenance work is not triggered by notifications of maintenance content based on the maintenance display device 100A.

[0320] In step S93, it is determined that the newly registered maintenance actual result data MD new If the maintenance plan ID is included (step S93: Yes), the maintenance effect determination unit 123A advances the process to step S94. On the other hand, it determines that the maintenance actual result data MD... new If the maintenance scheme ID is not included (step S93: No), the maintenance effect determination unit 123A advances the process to step S95.

[0321] Step S94 is the newly registered maintenance actual result data MD newThe processing for maintenance operations triggered by a notification of maintenance content from the maintenance display device 100A. Therefore, in step S94, the maintenance effect determination unit 123A transfers to processing to determine whether the maintenance operation triggered by the notification of maintenance content from the maintenance display device 100A is effective. Furthermore, the maintenance effect determination processing for maintenance triggered by the notification of maintenance content from the maintenance display device 100A is different from the processing described in the first embodiment above. Figure 18 The processing of the instructions is largely the same, so the instructions are omitted.

[0322] On the other hand, step S95 is to maintain the actual result data MD new This refers to the processing when the corresponding maintenance operation is not triggered by a notification of maintenance content based on the maintenance display device 100A. Therefore, the maintenance effect determination unit 123A shifts to processing to determine whether the maintenance operation, which is not triggered by the maintenance display device 100A, is effective. Furthermore, the maintenance effect determination processing for maintenance not triggered by a notification of maintenance content based on the maintenance display device 100A differs from the processing described in the first embodiment above. Figure 11 The processing of the instructions is largely the same, so the instructions are omitted.

[0323] As explained above, according to the maintenance display device 100A of the second embodiment, even when performing maintenance work that is not triggered by a notification of maintenance content based on the maintenance display device 100A, it is possible to appropriately register the actual maintenance result data MD. new In addition, utilizing Figure 22 The processing of the maintenance effect determination unit 123A described above can be performed in both the learning processing and the update processing.

[0324] The maintenance display device according to this embodiment generates equipment status diagnostic indicators before maintenance operations by establishing an associated equipment status diagnostic model based on actual production results data for the period from the time of maintenance operation where newly registered maintenance results data was performed to a previously given time, and the content of maintenance operations in the notification that triggered the maintenance operation that became the newly input maintenance results data. Furthermore, it generates equipment status diagnostic indicators after maintenance operations by establishing an associated equipment status diagnostic model based on actual production results data for the period from the time of maintenance operation where newly input maintenance results data was performed to a subsequently given time, and the content of maintenance operations in the notification that triggered the maintenance operation that became the newly input maintenance results data. Then, the maintenance effect determination unit calculates the difference between the equipment status diagnostic indicators before and after maintenance operations, and determines whether the maintenance operation was effective based on the magnitude of the difference.

[0325] (Third Implementation)

[0326] The third embodiment of this disclosure will now be described. Figure 23 This is a diagram illustrating the structure of the maintenance display device 100B according to the third embodiment. The maintenance display device 100B according to the third embodiment differs from the maintenance display device 100 according to the first embodiment described above in that the storage unit 110B of the server 10B also has an ineffective device status diagnostic model database 114, and the control unit 120B has a notification determination unit 122B, a maintenance effect determination unit 123B, and a device status diagnostic model generation unit 124B.

[0327] In the first embodiment described above, the equipment condition diagnostic model generation unit 124 generates a new equipment condition diagnostic model M using the maintenance actual results data MD that is determined to be effective. new

[0328] (Refer to Figure 13 In the third embodiment, the equipment condition diagnostic model generation unit 124B also uses the actual maintenance result data MD, which is determined to be ineffective, to generate a new equipment condition diagnostic model M. new .

[0329] Figure 24 This is a flowchart explaining the processing performed by the device condition diagnostic model generation unit 124B in the third embodiment. Additionally, using... Figure 24 The described processing can be performed in both learning and update processes.

[0330] In step S101, the equipment condition diagnosis model generation unit 124B reads the newly registered maintenance actual result data MD from the maintenance actual result database 113. new Here, the equipment condition diagnosis model generation unit 124B reads out the actual maintenance result data MD independently of the effect determination result based on the maintenance effect determination unit 123B. new .

[0331] In step S102, the equipment condition diagnosis model generation unit 124B reads the production actual results data list PL before the maintenance operation from the production actual results database 111. before .

[0332] In step S103, the equipment condition diagnostic model generation unit 124B uses the read maintenance actual result data MD new List of actual production results data PL before The actual production results data (PD) included are used to generate the equipment condition diagnosis model (M). new .

[0333] In step S104, the equipment condition diagnostic model generation unit 124B generates the newly generated equipment condition diagnostic model M. new The model generated based on the actual maintenance results data (MD) deemed ineffective is registered in the ineffective equipment condition diagnosis model database 114. Meanwhile, the equipment condition diagnosis model generation unit 124B generates a newly generated equipment condition diagnosis model M. new The model generated based on the actual maintenance results data (MD) that are determined to be effective is registered in the equipment condition diagnosis model database 112.

[0334] In this way, the equipment condition diagnosis model generation unit 124B not only generates an equipment condition diagnosis model M that utilizes the actual maintenance result data MD of maintenance that is determined to be effective, but also generates an equipment condition diagnosis model M that utilizes the actual maintenance result data MD of maintenance that is determined to be ineffective.

[0335] The generated equipment condition diagnostic model M is used to perform identification processing based on the equipment condition diagnostic unit 121 and the notification determination unit 122B. The processing performed by the equipment condition diagnostic unit 121 is similar to that described in the first embodiment above. Figure 15 The processing of the instructions is largely the same, so the instructions are omitted.

[0336] The following describes the process performed by the determination unit 122B in the identification process of the third embodiment. Figure 25 This is a flowchart for explaining the processing performed by the notification determination unit 122B in the third embodiment.

[0337] In step S111, the determination unit 122B is notified to use the consistency C generated by the equipment status diagnosis unit 121 to sum the consistency C for each maintenance group and generate a total value A. Furthermore, in the third embodiment, for each maintenance group, information (marks) indicating whether a maintenance operation is deemed effective is associated with the maintenance effectiveness determination unit 123B.

[0338] In step S112, the determination unit 122B is notified to generate a maintenance plan list ML, which is a list of maintenance teams arranged in descending order of total value A.

[0339] In step S113, the determination unit 122B is notified to determine whether each maintenance team included in the maintenance plan list ML is deemed effective. As described above, in the third embodiment, the equipment status diagnosis unit 121 establishes an association between each maintenance team and a marker indicating whether it is effective or not, therefore the determination unit 122B is notified to perform the processing of this step S113 with reference to this marker. The determination unit 122B is notified that for maintenance teams whose maintenance operations are deemed effective (step S113: Yes), the processing proceeds to step S114. On the other hand, the determination unit 122B is notified that for maintenance teams whose maintenance operations are deemed ineffective (step S113: No), the processing proceeds to step S117.

[0340] In step S114, the determination unit 122B is notified to determine whether the total value A for each maintenance team deemed effective is greater than the given warning threshold Th. f Even if there is only one total value A greater than the warning threshold Th... f In the case of the maintenance team (step S114: Yes), the notification determination unit 122B also advances the process to step S115. If there is no total value A greater than the warning threshold Th... f In the case of the maintenance team (step S115: no), notify the determination unit 122B to end the process.

[0341] In step S115, the determination unit 122B is notified to determine whether there is a total value A greater than a given anomaly threshold Th among the maintenance teams that have been determined to be effective. a The maintenance team. When the total value A exceeds the abnormal threshold Th... a In the case of a maintenance team (step S115: Yes), the determination unit 122B is notified to proceed to step S116. If the total value A is not greater than the abnormal threshold Th... a In the case of the maintenance team (step S115: no), notify the determination unit 122B to proceed to step S118.

[0342] In step S116, the notification determination unit 122B notifies that the total value A determined in step S114 is greater than the warning threshold Th. f The maintenance team is responsible for the maintenance tasks and issues an alarm to notify that an abnormality has occurred in the object winding device 200.

[0343] In step S117, the determination unit 122B is notified to determine whether the total value A for each maintenance group whose maintenance was determined to be ineffective is greater than the given ineffectiveness threshold Th. ie Ineffective threshold Th ie This is the minimum total value of the assumed ineffective notification. This applies when the total value A exceeds the ineffectiveness threshold Th. ieIn the case of a maintenance team (step S117: Yes), the determination unit 122B is notified to proceed to step S118. If there is no total value A greater than the ineffective threshold Th... ie In the case of the maintenance team (step S117: no), notify the determination unit 122B to end the process.

[0344] In step S118, the notification determination unit 122B notifies that the total value A determined in step S114 is greater than the warning threshold Th. f The maintenance content corresponding to the maintenance team. Simultaneously, the notification determination unit 122B notifies that in step S117, it was determined that the total value A is greater than the ineffective threshold Th. ie The maintenance content corresponding to the maintenance team.

[0345] With this structure, the maintenance display device 100B according to the third embodiment can not only notify the user of maintenance tasks that are assumed to improve the winding device 200, but also notify the user of maintenance tasks that were previously performed but were ineffective. This avoids the situation where the user repeatedly performs ineffective maintenance tasks, thus shortening the maintenance time and reducing the labor required for maintenance.

[0346] In the maintenance display device according to this embodiment, the equipment status diagnostic model generation unit generates a new equipment status diagnostic model based on actual production results data prior to the time point when a maintenance operation was determined to be ineffective, and actual maintenance results data related to that maintenance operation. The notification unit notifies the user of maintenance operations determined to be effective as effective maintenance operations, and notifies the user of maintenance operations that have been associated with the equipment status diagnostic model generated based on the actual maintenance results data related to maintenance operations determined to be ineffective as ineffective maintenance operations.

[0347] (Modified Example)

[0348] The embodiments described above with reference to the accompanying drawings are for illustrative purposes only and are not limited to this example. It is self-evident to those skilled in the art that various modifications or alterations can be conceived within the scope of the technical solution described, and these modifications are also understood to fall within the technical scope of this disclosure. Furthermore, the constituent elements of the above embodiments can be arbitrarily combined without departing from the spirit of the disclosure.

[0349] <Variation Example 1>

[0350] In the above-described embodiment, during the learning process, in the determination process of whether the maintenance operation is effective or not performed by the maintenance effect determination unit 123, the effectiveness is determined by whether the difference between the defect rate before and after the maintenance operation is greater than a given threshold (see reference). Figure 12A as well as Figure 12B ).

[0351] However, the maintenance effect determination unit 123 can also use other methods to determine whether the maintenance work is effective. Figure 26A as well as Figure 26B This is a diagram illustrating a variation of the method for determining the effectiveness of maintenance operations performed by the maintenance effectiveness determination unit 123 in the learning process.

[0352] exist Figure 26A as well as Figure 26B In the example shown, the failure rate before maintenance is not referenced, but rather based on the failure rate after maintenance, Nf. after Whether it has an effect is determined by whether it exceeds a given threshold (e.g., 20%). Figure 26A In the example shown, Nf after =0%, which is less than the given threshold of 20%, therefore it is considered effective. On the other hand, in Figure 26B In the example shown, Nf after =40%, which is greater than the given threshold of 20%, so it is determined to be ineffective.

[0353] Similarly, even during the update process, the maintenance effect determination unit 123 can use a method different from the above-described implementation to determine whether the maintenance operation is effective.

[0354] Furthermore, in the above-described embodiments, during the update process, in the process of determining whether the maintenance operation performed by the maintenance effect determination unit 123 is effective, the effectiveness is determined based on whether the difference in consistency before and after the maintenance operation is greater than a given threshold (see reference). Figure 19A as well as Figure 19B ).

[0355] Figure 27A as well as Figure 27B This is a diagram illustrating a modified example of the method for determining the effectiveness of maintenance operations performed by the maintenance effectiveness determination unit 123 during the update process.

[0356] exist Figure 27A as well as Figure 27B In the example shown, the consistency is not referenced before maintenance, but based on the consistency C after maintenance. after The effectiveness is determined by whether the value exceeds a given threshold (e.g., 0.30). Figure 27A In the example shown, C after =0.20, which is less than the given threshold of 0.30, therefore it is determined to be effective. On the other hand, in Figure 27B In the example shown, C after =0.90, which is greater than the given threshold of 0.30, so it is determined to have no effect.

[0357] <Variation Example 2>

[0358] In the above embodiment, the equipment condition diagnostic model generation unit 124 generates an equipment condition diagnostic model M. This equipment condition diagnostic model M is a learned model that has been completed, learning which maintenance operations are effective for which type of defect. The determination unit 122 then uses this model to determine whether maintenance operations should be performed. However, this disclosure is not limited to this, and it may also be based solely on the shape data of the produced wound body 204 (see reference). Figure 5 This is used to determine whether the winding body 204 is defective. Furthermore, if the winding body 204 is defective, a notification can be issued to allow maintenance work to be performed on any one of the multiple cores 206 that are the cause of the defect, or on the first bonding roller 205A or the second bonding roller 205B.

[0359] In this variation 2, the control unit of the maintenance display device performs control as follows: When the control unit acquires new shape data of the wound body, it determines whether the continuity of the positions of the two ends of the first sheet and the second sheet is parallel to the reference line. If it is determined to be parallel, the control unit does not notify that maintenance work should be performed. If it is determined to be non-parallel, the control unit determines whether only some of the multiple wound bodies wound on multiple cores are defective, or whether all of the multiple wound bodies wound on multiple cores are defective.

[0360] If only some of the multiple windings wound on multiple cores are defective, the control unit notifies that the cores with the defective windings should be maintained. On the other hand, if all the multiple windings wound on multiple cores are defective, the control unit notifies that at least one of the first bonding roller 205A and the second bonding roller 205B should be maintained.

[0361] In Modification 2, the control unit can also change the notification method when the continuous tilt angle at both ends of the first and second sheets is below a given threshold and when it is above the threshold. With this structure, even without generating an equipment condition diagnostic model M, it is possible to determine which of the multiple cores 206, or the first bonding roller 205A or the second bonding roller 205B, should be maintained and to issue a notification. However, compared to Modification 2, the accuracy of determining the cause of the defect is higher in Embodiments 1-3 described above; therefore, for the purposes of this disclosure, Embodiments 1-3 described above are more suitable than Modification 2.

[0362] <Variation Example 3>

[0363] In the above embodiments, for illustrative purposes, the structure of the maintenance display devices 100 (100A, 100B) including a storage unit 110 (110B), a control unit 120 (120A, 120B), and a notification unit 130 has been described, but this disclosure is not limited thereto. Although the above embodiments have been described, in this disclosure, the storage unit and the control unit can be configured to communicate with each other, or they can be configured separately and arranged in separate positions. Furthermore, the notification unit can be included in the production apparatus or installed outside the production apparatus. In addition, the notification unit can be connected to the storage unit and the control unit via a network or directly connected.

[0364] Thus, in the maintenance display device disclosed herein, the storage unit, control unit, and notification unit can each be an independent, separate device that operates independently. Furthermore, the storage unit, control unit, and notification unit only need to be able to communicate with each other; there are no particular limitations on their location. Alternatively, the notification device could be located, for example, in a factory where the production equipment is located, while the storage unit and control unit could be contained, for example, in a so-called cloud server configured in the cloud.

[0365] Furthermore, in the above-described embodiments, the control units 120 (120A, 120B) perform all of the learning process, the updating process, and the identification process. The learning process refers to the process of generating the equipment condition diagnostic model M, and the updating process refers to the process of updating the equipment condition diagnostic model M. The identification process refers to the process of using the equipment condition diagnostic model M to identify whether any abnormalities or signs of abnormalities have occurred in the newly produced plurality of wound bodies 204. Moreover, in the identification process, the control units 120 (120A, 120B) control the notification unit 130 to perform notification processing. However, this disclosure is not limited to this.

[0366] For example, it can be configured such that the control unit only performs learning or update processing, while the notification unit receives the device status diagnostic model from the control unit and uses the received model for identification processing. With this structure, the increase in communication volume between the control unit and the notification unit can be suppressed, and even when multiple notification units are connected to the control unit, the load of identification processing can be distributed among the notification units, thus preventing processing delays caused by processing being concentrated in the control unit.

[0367] According to this disclosure, it is possible to detect signs of anomalies in a device.

[0368] Industrial availability

[0369] This disclosure is useful in maintenance display devices that display information related to the maintenance of production equipment.

Claims

1. A method for generating a learned model, wherein the learned model is used for the maintenance of a winding device, the winding device comprising: The first supply unit supplies the first electrode sheet; The second supply mechanism supplies the second electrode sheet; The first bonding roller is disposed on the side of the first electrode sheet; The second bonding roller is disposed on the side of the second electrode sheet and is paired with the first bonding roller to bond the first electrode sheet and the second electrode sheet together; Volume 1 core; Volume 2 core; A driving mechanism moves the first core to a given winding position, overlaps and winds the first electrode sheet and the second electrode sheet around the first core, moves the second core to the given winding position, and overlaps and winds the first electrode sheet and the second electrode sheet around the second core. and The sensor reads the first end face of the first electrode and the second end face of the second electrode along the radial direction of the first winding body, in which the first electrode and the second electrode are wound in multiple overlapping turns on the first core; and reads the third end face of the first electrode and the fourth end face of the second electrode along the radial direction of the second winding body, in which the first electrode and the second electrode are wound in multiple overlapping turns on the second core. The method for generating the learned model includes the following steps: The sensor acquires a first set of data indicating the position of the first end face read along the radial direction of the first winding body, a second set of data indicating the position of the second end face read along the radial direction of the first winding body, a third set of data indicating the position of the third end face read along the radial direction of the second winding body, and a fourth set of data indicating the position of the fourth end face read along the radial direction of the second winding body. If the position of the third end face represented by the third set of data is continuous and the position of the fourth end face represented by the fourth set of data is inclined relative to the baseline, it is determined that the second winding body is defective. The first learning completed model is generated using the third set of data and the fourth set of data to output information indicating that the cause of the defect is the second winding core. A second learned model is generated using the first set of data, the second set of data, the third set of data, and the fourth set of data. This second learned model takes the fifth set of data and the sixth set of data as input in the following case and outputs information indicating that the third winding is defective and the cause of the defect is the first core. The following case is: the fifth set of data, which indicates the position of the fifth end face of the first electrode, is obtained from the sensor along the radial direction of the third winding, which has the first electrode and the second electrode wrapped in multiple overlapping turns on the first core, and the sixth set of data, which indicates the position of the sixth end face of the second electrode, is obtained along the radial direction of the third winding. The continuity of the position of the fifth end face indicated by the fifth set of data and the continuity of the position of the sixth end face indicated by the sixth set of data are tilted relative to the baseline, which is the case where the third winding is defective.

2. The method for generating a learned model according to claim 1, wherein, The first electrode plate is the positive electrode plate of the battery, and the second electrode plate is the negative electrode plate of the battery.

3. The method for generating a learned model according to claim 1, wherein, The first electrode plate is the negative electrode plate of the battery, and the second electrode plate is the positive electrode plate of the battery.

4. The method for generating a learned model according to any one of claims 1 to 3, wherein, The method for generating the learned model also includes the following steps: Using the first set of data, the second set of data, the third set of data, and the fourth set of data, a first dataset is created by combining the first roll core with the first set of data and the second set of data; a second dataset is created by combining the second roll core with the third set of data and the fourth set of data; a first swap dataset is created by combining the first roll core with the third set of data and the fourth set of data; and a second swap dataset is created by combining the second roll core with the first set of data and the second set of data. The second learned model is generated using the first dataset, the second dataset, the first swap dataset, and the second swap dataset. This second learned model is used to output information indicating that the third winding is defective and the reason for the defect is the first winding core, given the input of the fifth set of data and the sixth set of data.

5. An information output device for displaying information related to the maintenance of a winding device, the winding device comprising: The first supply unit supplies the first electrode sheet; The second supply mechanism supplies the second electrode sheet; The first bonding roller is disposed on the side of the first electrode sheet; The second bonding roller is disposed on the side of the second electrode sheet and is paired with the first bonding roller to bond the first electrode sheet and the second electrode sheet together; Volume 1 core; Volume 2 core; A driving mechanism moves the first core to a given winding position, overlaps and winds the first electrode sheet and the second electrode sheet around the first core, moves the second core to the given winding position, and overlaps and winds the first electrode sheet and the second electrode sheet around the second core. and The sensor reads the first end face of the first electrode and the second end face of the second electrode along the radial direction of the first winding body, in which the first electrode and the second electrode are wound in multiple overlapping turns on the first core; and reads the third end face of the first electrode and the fourth end face of the second electrode along the radial direction of the second winding body, in which the first electrode and the second electrode are wound in multiple overlapping turns on the second core. The device for outputting information includes: The acquisition unit acquires from the sensor a first set of data indicating the position of the first end face read along the radial direction of the first winding body, a second set of data indicating the position of the second end face read along the radial direction of the first winding body, a third set of data indicating the position of the third end face read along the radial direction of the second winding body, and a fourth set of data indicating the position of the fourth end face read along the radial direction of the second winding body; and The model generation unit determines that the second winding is defective when the positions of the third end face represented by the third set of data and the fourth end face represented by the fourth set of data are continuously tilted relative to the baseline. It then uses the third and fourth sets of data to generate a first learned model that outputs information indicating the defect is due to the second winding core. A second learned model is generated using the first, second, third, and fourth sets of data. This second learned model takes the fifth and sixth sets of data (under the following conditions) as input and includes information indicating the third winding is defective. Furthermore, the reason for the defect is the information of the first winding core as the output. The above situation is: the sensor obtains the fifth set of data indicating the position of the fifth end face of the first electrode sheet along the radial direction of the third winding body, which has the first electrode sheet and the second electrode sheet wrapped in multiple overlapping turns on the first winding core, and obtains the sixth set of data indicating the position of the sixth end face of the second electrode sheet along the radial direction of the third winding body. The continuity of the position of the fifth end face indicated by the fifth set of data and the continuity of the position of the sixth end face indicated by the sixth set of data are tilted relative to the reference line, and the third winding body is judged to be defective.

6. A computer-readable recording medium recording a program executed by a computer, the computer generating a learned model for the maintenance of a winding device, the winding device comprising: The first supply unit supplies the first electrode sheet; The second supply mechanism supplies the second electrode sheet; The first bonding roller is disposed on the side of the first electrode sheet; The second bonding roller is disposed on the side of the second electrode sheet and is paired with the first bonding roller to bond the first electrode sheet and the second electrode sheet together; Volume 1 core; Volume 2 core; A driving mechanism moves the first core to a given winding position, overlaps and winds the first electrode sheet and the second electrode sheet around the first core, moves the second core to the given winding position, and overlaps and winds the first electrode sheet and the second electrode sheet around the second core. and The sensor reads the first end face of the first electrode and the second end face of the second electrode along the radial direction of the first winding body, in which the first electrode and the second electrode are wound in multiple overlapping turns on the first core; and reads the third end face of the first electrode and the fourth end face of the second electrode along the radial direction of the second winding body, in which the first electrode and the second electrode are wound in multiple overlapping turns on the second core. The program causes the computer to perform the following process: Acquire from the sensor a first set of data indicating the position of the first end face read along the radial direction of the first winding body, a second set of data indicating the position of the second end face read along the radial direction of the first winding body, a third set of data indicating the position of the third end face read along the radial direction of the second winding body, and a fourth set of data indicating the position of the fourth end face read along the radial direction of the second winding body; and If the positions of the third end face represented by the third set of data and the positions of the fourth end face represented by the fourth set of data are both inclined relative to the baseline, then the second winding is determined to be defective. A first learning-completed model is then generated using the third and fourth sets of data to output information indicating that the defect is due to the second winding core. A second learned model is generated using the first set of data, the second set of data, the third set of data, and the fourth set of data. This second learned model takes the fifth set of data and the sixth set of data as input in the following case and outputs information indicating that the third winding is defective and the cause of the defect is the first core. The following case is: the fifth set of data, which indicates the position of the fifth end face of the first electrode, is obtained from the sensor along the radial direction of the third winding, which has the first electrode and the second electrode wrapped in multiple overlapping turns on the first core, and the sixth set of data, which indicates the position of the sixth end face of the second electrode, is obtained along the radial direction of the third winding. The continuity of the position of the fifth end face indicated by the fifth set of data and the continuity of the position of the sixth end face indicated by the sixth set of data are tilted relative to the baseline, which is the case where the third winding is defective.

Images