Work instruction system, work instruction method, and production method for recycled product

Abstract

According to the present invention, a work instruction system that outputs work instructions for remanufacturing comprises a storage unit that stores a first learning model that has been trained by machine learning on the correspondence between the states of devices and work instructions for work to be performed on devices in the relevant states and device state information that indicates the state of a target device that is to be remanufactured, a control unit that inputs the device state information for the target device as acquired from the storage unit into the first learning model to acquire information that indicates work instructions for work to be performed on the target device, and a display unit that outputs work instructions for the target device on the basis of the information acquired from the first learning model by the control unit.

Description

Work Instruction System, Work Instruction Method, and Method for Manufacturing Remanufactured Products

[0001] The present disclosure relates to a technique for giving work instructions in consideration of the individual state of a product when remanufacturing industrial products.

[0002] Generally, industrial products are manufactured by forming parts from raw materials such as metal and oil and assembling those parts. A large amount of energy is required in the process up to that point of productization, and the more new products are produced, the more the CO ２ emission amount increases. Therefore, in recent years, remanufacturing has attracted attention as an effort to contribute to resource conservation and energy saving. Remanufacturing is an effort to collect used products and parts and regenerate them as parts or products again. However, processes from collection to disassembly, cleaning, repair, inspection, reassembly, and final inspection require labor and capital investment in equipment, resulting in high costs.

[0003] To solve this problem, in Patent Document 1, in the recycling and reuse process for recycled office equipment such as copiers, different disassembly, cleaning, and reassembly operations for each unit are presented to workers according to the equipment characteristics and usage history / background in the market. This enables accurate and rapid work, improves productivity, and reduces the recycling cost.

[0004] In Patent Document 2, regarding rebuilt products of defective products in product assembly, the cause of the defect is analyzed from product information and product inspection information, and the work procedure with the highest evaluation in the instruction manual linked to the cause is presented using a score based on rules. This can increase the success rate of product rebuilding without being influenced by the ability of the analyst.

[0005] Japanese Patent Application Laid-Open No. 10-216689, Japanese Patent Application Laid-Open No. 2011-076488

[0006] Conventional methods classify product conditions into several patterns based on rules and present work instructions to workers according to those patterns. However, because they are rule-based, it is difficult to provide work procedures that take into account the specific circumstances of each product. Therefore, there were problems such as inconsistencies and waste due to unnecessary work, and the output of inappropriate work instructions. In order to further improve the efficiency of remanufacturing industrial products, there was a need for a method that could provide appropriate work instructions that take into account the specific circumstances of each product.

[0007] This disclosure is made in light of the above-mentioned circumstances and aims to provide technology for presenting appropriate work instructions according to the condition of the product.

[0008] This disclosure includes a work instruction system for outputting work instructions for remanufacturing, comprising: a first learning model that has learned by machine learning the correspondence between the state of a device and work instructions for work to be performed on the device in that state; a storage unit that stores device state information indicating the state of a target device to be remanufactured; a control unit that acquires information indicating work instructions for work to be performed on the target device by inputting the device state information of the target device acquired from the storage unit into the first learning model; and a display unit that outputs work instructions for the target device based on the information acquired from the first learning model by the control unit.

[0009] According to this disclosure, appropriate work instructions can be provided depending on the condition of the product.

[0010] Figure 1 shows an example of a functional block of the work instruction system. Figure 2 shows an example of the hardware configuration of the work instruction system. Figure 3 shows an example of the inspection information acquisition unit. Figure 4 shows an example of the process of generating and updating a learning model for determining work instructions. Figure 5 shows an example of the process of acquiring work instructions using the learning model in Figure 4. Figure 6 shows an example of the process of generating and updating a learning model for determining soiling and wear. Figure 7 shows an example of the process of determining the soiling and wear status using the learning model in Figure 6. Figure 8 shows an example of the process of generating and updating a learning model for proposing additional work and updating procedure manuals. Figure 9 shows an example of the process of determining additional work and updating procedure manuals using the learning model in Figure 8. Figure 10 shows an example of a screen displayed when issuing work instructions. Figure 11 shows an example of a screen displayed when issuing cleaning instructions. Figure 12 shows an example of a screen that displays the work plan and the number of replacement parts. Figure 13 shows an example of a screen that proposes changing the location of parts. Figure 14 is a flowchart showing the process of generating and updating the learning model in Figure 4. Figure 15 is a flowchart showing the process of generating and updating the learning model in Figure 6. Figure 16 is a flowchart showing the process of generating and updating the learning model in Figure 8. Figure 17 is a flowchart showing the processing procedure for work instructions. Figure 18 is a diagram showing an example of a screen that displays the number of reusable parts.

[0011] The embodiments of this disclosure will be described below. Note that these embodiments are merely examples of the disclosure.

[0012] [Functional Block Configuration] Figure 1 is a schematic functional block diagram illustrating a work instruction system corresponding to remanufacturing according to one embodiment.

[0013] As shown in Figure 1, the work instruction system 1 according to one embodiment is a system that provides appropriate instructions for remanufacturing work on the device 100 according to the state of the device 100. The work instruction system 1 of this embodiment mainly comprises a sensor 10 for performing inspection of the device 100 and an information processing device 200.

[0014] Device 100 is an industrial product that is to be remanufactured. Device 100 can be any type of device, but for example, it may be office equipment such as a multifunction printer or copier. Device 100 is equipped with (built into) a usage history information storage unit 101 that manages the device's usage history. For example, if device 100 is a copier, the usage history information storage unit 101 stores and manages information such as the number of pages printed, operating time, and parts replacement history as usage history information.

[0015] Sensor 10 is a device for acquiring the status of the apparatus 100, and an appropriate sensor is used depending on the type of apparatus 100 and its intended use. Examples of sensors 10 include imaging sensors (cameras), ultrasonic sensors, X-ray sensors, force sensors, torque sensors, and vibration sensors. Multiple sensors may be used in combination.

[0016] The device 100 and the information processing device 200 are preferably connected via wired or wireless communication, and the information processing device 200 is capable of collecting data from the usage history information storage unit 101 installed in the device 100. If the device 100 and the information processing device 200 are not capable of communication, the usage history information of the device 100 can be input to the information processing device 200 via a storage medium or by user input. The sensor 10 and the information processing device 200 are also preferably connected via wired or wireless communication, and the information processing device 200 is capable of collecting data from the sensor 10. If the sensor 10 and the information processing device 200 are not capable of communication, the information acquired by the sensor 10 can be input to the information processing device 200 via a storage medium or by user input.

[0017] The information processing device 200 mainly comprises a control unit 210, a storage unit 220, a display unit 230, and an input unit 240.

[0018] Each of the functional blocks 211 to 216 shown in the control unit 210 is realized by the CPU, GPU, and other processors of the computer constituting the information processing device 200 controlling the operation of each part of the device according to their respective programs. Details of each functional block 211 to 216 will be described later.

[0019] The memory unit 220 is a data storage unit that stores various data necessary for executing the work instruction presentation process. The memory unit 220 includes a device state storage unit 221, a work history information storage unit 222, a work instruction information storage unit 223, and a learning model storage unit 224. These sub-storage units of the memory unit 220 are configured by appropriately allocating them to the storage areas of storage devices such as hard disk drives, RAM, and ROM.

[0020] The display unit 230 and the input unit 240 are user interfaces provided by the information processing device 200. The display unit 230 can use a display device such as a liquid crystal display, an organic EL display, or a head-mounted display. The input unit 240 can use an input device such as a keyboard, a jog dial, a mouse, a pointing device, or a voice input device.

[0021] The following describes the various functional units provided by the control unit 210.

[0022] The usage history information acquisition unit 211 acquires usage history information from the usage history information storage unit 101 of the device 100 and stores it in the device state storage unit 221. The usage history information is information that shows the usage history of the device 100. For example, if the device 100 is a copier, the usage history information acquisition unit 211 collects and stores usage history information such as the number of printed pages, operating time, and internal temperature managed by the device 100.

[0023] The inspection information acquisition unit 212 acquires inspection information from the sensor 10, which is the result of the inspection of the device 100 performed using the sensor 10, and stores it in the device state storage unit 221. For example, when the device 100 is imaged by a camera and an external inspection is performed, the inspection information acquisition unit 212 acquires image data of the device 100 from the camera and stores it in the device state storage unit 221. The inspection information acquisition unit 212 may store the data acquired from the sensor 10 directly in the device state storage unit 221, or it may store the data acquired from the sensor 10 after processing it.

[0024] The learning model generation and update unit 213 generates a learning model using the data stored in the device state storage unit 221, the work history information storage unit 222, and the work instruction information storage unit 223, and stores it in the learning model storage unit 224. In this embodiment, three learning models are used: a learning model for determining work instructions and status (first learning model), a learning model for determining soiling and wear status (second learning model), and a learning model for determining additional work and updating procedure manuals (third learning model). The learning model for determining work instructions and status is a model that has learned by machine learning the correspondence between the state of the device and the work instructions for the work to be performed on the device in that state. The learning model for determining soiling and wear status is a model that has learned by machine learning the correspondence between the state of the device and the location and content of soiling and wear on the device in that state. The learning model for determining additional work and updating procedure manuals is a model that has learned by machine learning the correspondence between the state of the device and the additional work required for the device in that state. The machine learning algorithm may be, for example, an inference process using a neural network, or it may be a well-known machine learning algorithm such as a support vector machine or a Gaussian mixture model.

[0025] The work instruction / status determination unit 214 inputs the data stored in the device state storage unit 221 into a learning model (first learning model) stored in the learning model storage unit 224 for determining work instructions and status, and obtains an output result. Based on the output result of the learning model, the work instruction / status determination unit 214 determines the work instructions to be performed on the device 100, and also determines the progress and status of the work.

[0026] Furthermore, the work instruction / status determination unit 214 can also determine the completion of a task using a learning model (first learning model) for determining work instructions and status. For example, the work instruction / status determination unit 214 inputs the device status of the device 100 (such as inspection information) after the work has been performed according to the work instructions into the learning model. If the learning model outputs a work instruction for the task to be performed after the previously instructed task, the unit may determine that the previously instructed task has been completed. If the learning model outputs the same work instruction as before, the work instruction / status determination unit 214 may determine that the previously instructed task has not been completed (the task was omitted or there was a mistake).

[0027] The contamination / wear status determination unit 215 inputs the data stored in the device state storage unit 221 into a learning model (second learning model) stored in the learning model storage unit 224 for determining contamination / wear status, and obtains an output result. The contamination / wear status determination unit 215 determines the contamination / wear status from the output result.

[0028] The additional work / procedure manual update determination unit 216 inputs the data stored in the device state storage unit 221 into a learning model (third learning model) stored in the learning model storage unit 224 for determining whether or not to update the additional work / procedure manual, and obtains an output result. The additional work / procedure manual update determination unit 216 determines whether or not to update the additional work / procedure manual from the output result.

[0029] The work instruction display unit 231 included in the display unit 230 displays the results determined by the work instruction / status determination unit 214, the soiling / wear status determination unit 215, and the additional work / procedure manual update determination unit 216 on the display unit 230.

[0030] [Hardware Configuration] Figure 2 schematically shows an example of the hardware configuration of the work instruction system corresponding to the remanufacturing embodiment. The work instruction system 1 corresponding to remanufacturing, as shown in Figure 2, may include PC hardware equipped with a CPU 401 as the main control means, a ROM 402 as a storage device, and a RAM 403. The ROM 402 can store information such as processing programs and work instruction judgment algorithms for realizing the work instruction method described later. The RAM 403 is used as the work area of ​​the CPU 401 when executing the control procedure. An external storage device 406 is also connected to the control system. The external storage device 406 consists of an HDD, an SSD, or an external storage device of another network-mounted system.

[0031] The processing program for implementing the work instruction method of this embodiment, described later, can be stored in an external storage device 406 or a storage unit such as ROM 402 (e.g., EEPROM area). The processing program for implementing the work instruction method may be supplied via a network such as the Internet. In that case, the processing program is acquired via a network interface (NIF) 407 and stored in each of the above-mentioned storage units. Furthermore, update programs can be acquired via the network interface (NIF) 407, and the processing programs stored in each storage unit can be updated with new (different) programs. Alternatively, the processing program for implementing the work instruction method may be supplied offline. For example, it can be supplied to each of the above-mentioned storage units and its contents updated via various storage means such as magnetic disks, optical disks, and flash memory, and a drive device for them. Various storage means, storage units, or storage devices in a state in which the processing program for implementing the work instruction method is stored constitute a computer-readable recording medium that non-temporarily stores the work instruction procedure of this disclosure.

[0032] The sensor 10 shown in Figure 1 is connected to the CPU 401. In Figure 2, for the sake of simplicity, the sensor 10 is shown as being directly connected to the CPU 401, but it may also be connected via, for example, IEEE 488 (so-called GPIB). Alternatively, the sensor 10 may be connected to the CPU 401 via network interface 407 and network 408.

[0033] The network interface (NIF) 407 can be configured using communication standards such as wired communication, IEEE 802.3, or wireless communication, IEEE 802.11, 802.15. The CPU 401 can communicate with other devices 100 and 300 via the network interface 407. For example, if the object to be remanufactured is a copier, the other device may be the copier 100, or it may be an online management server 300 set up to manage the copier's usage history, failure information, maintenance information, etc.

[0034] In the example shown in Figure 2, an input device 404 and a display device 405, corresponding to the input unit 240 and display unit 230 shown in Figure 1, are connected to the information processing device 200 as a UI device (user interface device). The input device 404 can be configured as a terminal such as a handheld terminal, or a device such as a keyboard (KB), jog dial, mouse, pointing device (PD), or voice input device (or a control terminal equipped with these). The display device 405 can display information related to the processing performed by the work instruction display unit 231, the learning model generation / update unit 213, the work instruction / status judgment unit 214, etc., on a display screen. The display device 405 can be, for example, a liquid crystal display, an organic EL display, or a head-mounted display.

[0035] [Regarding the method of giving work instructions] Next, the method of giving work instructions used in this embodiment will be explained.

[0036] As shown in Figure 3, the inspection information acquisition unit 212 acquires inspection information of the remanufacturing target device 100 (a copier in the example of Figure 3) using various sensors 10. For example, inspections that can identify the condition of contamination and wear of the device 100 include visual inspection, ultrasonic inspection, and X-ray inspection. Specific inspection information acquired through these inspections includes image information for visual inspection, reflection intensity information for ultrasonic inspection, and X-ray image information for X-ray inspection. The inspection information may be raw data (original data) acquired by the various sensors 10, or it may be data processed by the signal processing unit of the sensor 10.

[0037] Figure 4 shows a schematic diagram of the process for generating and updating a learning model for determining work instructions. The learning model 44 for determining work instructions is generated and updated, for example, by supervised learning using a neural network. The data used for learning consists of three types: device status information 41 (usage history information, inspection information), work history information 42, and device status information after reassembly 43 (inspection information after reassembly, repair / failure information). Device status information 41 is information indicating the state of the device 100 before remanufacturing (before disassembly), and device status information 43 is information indicating the state of the device 100 after remanufacturing (after reassembly).

[0038] Let's take a photocopier as an example to explain the training data in detail. The training data used includes device status information 41, work history information 42, and device status information 43 after reassembly, all collected from a large number of device samples.

[0039] The usage history information within the device status information 41 includes information such as the number of printed pages, operating time, and internal device temperature, which is stored in the usage history information storage unit 101. The inspection information within the device status information 41 includes image information from visual inspection, reflection intensity information from ultrasound inspection, and X-ray image information from X-ray inspection, which are stored in the device status storage unit 221.

[0040] The work history information 42 is information such as the procedure of the work and the type and number of parts replaced, which is stored in the work history information storage unit 222. In this case, the work procedure consists of the name of the procedure manual and the order of the work, which are the smallest units of work instructions stored in the work instruction information storage unit 223.

[0041] Of the device status information 43 after reassembly, the inspection information includes image information from visual inspection, reflection intensity information from ultrasonic inspection, and X-ray image information from X-ray inspection, all stored in the device status memory unit 221. Of the device status information 43 after reassembly, the repair / failure information is information about repairs or failures that occurred after the reassembled device 100 was shipped. The learning model generation / update unit 213 generates training data from the device status information 43 after reassembly. For example, based on the inspection pass rate obtained from the inspection information after reassembly and the prognosis of the failure (duration and cause) obtained from the repair / failure information, the unit sorts the work into correct and incorrect based on a certain level.

[0042] Figure 5 shows a schematic diagram of the input and output when outputting work instructions using the learning model 44 learned by the generation and update process shown in Figure 4. The work instruction / status judgment unit 214 inputs device status information 51 (usage history information, inspection information) and work instruction information 52 (procedure manual information which is the smallest unit of work instructions) that defines the work procedure into the learning model 44 and outputs work instructions 53. In order to output the overall plan, a work plan is output based on the usage history information even for areas that cannot be judged from the inspection information. At that time, the work instructions and work plan are output in the same format and are distinguished by the information used at the time of judgment.

[0043] Taking a copying machine as an example, the determination process of work instructions will be specifically described. Among the device state information 51, the usage history information is the number of printed sheets, operating time, etc. stored in the usage history information storage unit 101. Among the device state information 51, the inspection information is the image information of appearance inspection, reflection intensity information of ultrasonic inspection, X-ray image information of X-ray inspection, etc. stored in the device state storage unit 221. At this time, the usage history information and inspection information, which are the input data to the learning model 44, may be data in the same format as the usage history information and inspection information of the learning data used for the generation and update of the learning model 44. If the format of the input data is different from the format of the learning data, the input data may be pre-processed and converted into data in the same format as the learning data, and then input to the learning model 44. The work instruction information is the procedure manual information that is the minimum unit of the work instruction stored in the work instruction information storage unit 223, and includes the procedure manual name, information for determining whether it is disassembly or cleaning, etc. The work instruction / situation determination unit 214 determines the work instruction by the learning model 44 based on these information. The work instruction / situation determination unit 214 acquires and outputs information consisting of the procedure manual name, work order number, and whether inspection information is used from the learning model 44 for work instruction. At this time, when the work order numbers are the same, the procedures are parallelizable.

[0044] FIG. 6 shows a schematic diagram of a process for generating and updating a learning model for determining dirt and wear. The learning model 63 for determining dirt and wear is generated and updated, for example, by supervised learning using a neural network. The data used for learning includes two types: device state information 61 (usage history information, inspection information) for determining dirt and wear and dirt and wear state information 62 as teacher data.

[0045] Taking a copying machine as an example, the learning data will be specifically described. As the learning data, device state information 61 and dirt and wear state information 62 collected for a number of device samples are used.

[0046] Among the device status information 61, the usage history information is the number of printed sheets, operating time, usage status for each color, etc., stored in the usage history information storage unit 101. The inspection information among the device status information 61 is the image information of the appearance inspection, the reflection intensity information of the ultrasonic inspection, the X-ray image information of the X-ray inspection, etc., stored in the device status storage unit 221. The dirt and wear status information 62 may use, for example, part replacement information, cleaning work information, image information indicating the dirt and wear locations, reflection intensity information and X-ray image information indicating the wear locations, and the actual dirt and wear content information associated with these information. The dirt and wear status information 62 is, for example, information manually input by an operator and is used as correct data. The learning model generation / updating unit 213 inputs the device status information 61 and learns / updates the learning model 63 that outputs the dirt and wear status information 62. The output of the learning model 63 is, for example, information indicating the locations and contents of dirt and wear (which may include the types and degrees of dirt and wear).

[0047] FIG. 7 shows a schematic diagram of the input and output when performing dirt and wear determination using the learning model 63 learned by the generation / updating process shown in FIG. 6. The dirt and wear status determination unit 215 inputs the device status information 71 (usage history information, inspection information) for dirt and wear determination into the learning model 63 and outputs the dirt and wear determination information 72 and the dirt and wear location information 73. Also, for the range that cannot be determined by the inspection information, the dirt and wear determination information 72 and the dirt and wear location information 73 are output based on the usage history information.

[0048] The process for determining the condition of soiling and wear will be explained in detail using a photocopier as an example. Of the device status information 71, the usage history information is the number of printed pages, operating time, usage status for each color, etc., stored in the usage history information storage unit 101. Of the device status information 71, the inspection information is the image information of the visual inspection, the reflection intensity information of the ultrasonic inspection, the X-ray image information of the X-ray inspection, etc., stored in the device status storage unit 221. In this case, the usage history information and inspection information, which are input data to the learning model 63, should be in the same format as the learning data, which are the usage history information and inspection information, used to generate and update the learning model 63. If the format of the input data is different from the format of the learning data, the input data may be preprocessed and converted into data in the same format as the learning data before being input to the learning model 63. The soiling and wear condition determination unit 215 uses this information to determine the soiling and wear condition using the learning model 63. The outputted damage and wear status information includes, for example, damage and wear judgment information 72 such as the type of damage and wear, and image information that specifically shows the damaged or worn areas.

[0049] For example, around the toner cartridge of a copier, the type and extent of contamination can be identified from the external image acquired as inspection information. If there is a metallic sheen, it is metal powder; if it is black, it is toner; if it is black and solid, it is rubber fragments, etc. Then, the deterioration status of parts and the possibility of toner leakage can be determined from the usage status and operating time for each color in the usage history information. Using this information, the learning model 63 determines the type and location of the contamination, and if it is dirt, the cleaning method (whether disassembly is necessary, whether to use air, Kimwipes, ethanol, etc.), and notifies the operator. Note that the notification of the type of contamination / wear and the location of contamination / wear may be output in a format other than the format exemplified here; for example, the location of contamination / wear may be represented in text.

[0050] Figure 8 shows a schematic diagram of the process for generating and updating a learning model to propose updates to additional work procedures and manuals. The learning model 83 for proposing updates to additional work procedures and manuals is generated and updated, for example, by supervised learning using a neural network. The data used for learning consists of two types: device status information 81 (usage history information, inspection information) and work history information 82 which serves as training data.

[0051] Let's take a photocopier as an example to explain the training data in detail. As training data, device status information 81 and work history information 82 collected from a large number of device samples are used.

[0052] Of the device status information 81, the usage history information includes the number of printed pages, operating time, and usage status for each color, which are stored in the usage history information storage unit 101. Of the device status information 81, the inspection information includes image information for visual inspection, reflection intensity information for ultrasonic inspection, and X-ray image information for X-ray inspection, which are stored in the device status storage unit 221. The work history information 82 consists of the procedure manual name and additional work information. The procedure manual name indicates the procedure manual used when performing the work, and the additional work information is information stored as text when work not covered in the procedure manual is performed. In this case, the content of the additional work may be registered by the worker as text, or the additional work may be estimated and automatically registered using image recognition technology such as skeletal extraction and object recognition from video data of the additional work.

[0053] Figure 9 shows a schematic diagram of the input and output when proposing additional work and updating the procedure manual using the learning model 83 learned by the generation and update process shown in Figure 8. The additional work / procedure manual update determination unit 216 takes device status information 91 (usage history information, inspection information) and work instruction information 92 as input and outputs additional work / procedure manual update proposal information.

[0054] The process for determining additional work and updating procedure manuals will be explained in detail using a photocopier as an example. The usage history information among the device status information 91 includes the number of printed pages, operating time, usage status for each color, etc., stored in the usage history information storage unit 101. The inspection information among the device status information 91 includes image information for visual inspection, reflection intensity information for ultrasonic inspection, X-ray image information for X-ray inspection, etc., stored in the device status storage unit 221. In this case, the usage history information and inspection information, which are input data to the learning model 83, should be in the same format as the learning data, which are the usage history information and inspection information, used to generate and update the learning model 83. If the format of the input data differs from the format of the learning data, the input data may be preprocessed and converted into data in the same format as the learning data before being input to the learning model 83. The work instruction information 92 uses the procedure manual name and information on the order of work. The additional work / procedure manual update determination unit 216 uses this information to determine whether or not to propose additional work / procedure manual updates using the learning model 83. The outputted additional work / procedure update information includes, for example, the name of the procedure for which additional work / procedure update is proposed, and the details of the additional work.

[0055] Figure 10 shows an example of a work instruction screen in this embodiment. The work instruction screen 1001 is the screen displayed on the display unit 230 when a work instruction is given. The work instruction display unit 231 displays the work instruction screen 1001 as shown in Figure 10 based on information acquired from the learning models 44 and 83. This display content is generated by combining procedure manual information, which is the smallest unit of a work instruction, stored in the work instruction information storage unit 223.

[0056] If the work instructions include cleaning instructions, it is preferable to display work instructions for the soiling or wear of the target device 100 along with information indicating the location of the soiling or wear. For example, as shown in the work instruction screen 1101 of Figure 11, the actual image of the target device 100 captured during inspection, or marks (graphics, etc.) indicating the soiling or wear in the image, may be displayed as an image 1102 of the specific soiling or wear location. In addition, each procedure may include a checkbox 1103 to prevent work from being missed. The control unit 210 may use the checking of the checkbox 1103 as a trigger for determining that the work is complete.

[0057] Figure 12 shows an example of a work plan screen in this embodiment. The work plan screen 1201 includes an overall plan screen 1202 that represents the overall flow of all tasks to be performed on the device 100 in a flowchart in which the procedure manuals corresponding to each task are arranged in the order of execution. The overall plan screen 1202 presents, for example, the names of the procedure manuals, which are the smallest units of work instructions, as a flowchart, as shown in Figure 12. This flowchart is created based on the work sequence number output in the work instruction, and tasks with the same number are indicated on the flowchart so that they can be identified as tasks that can be processed in parallel. In the example in Figure 12, procedures A, B, C, D, G, and L, which cannot be processed in parallel, are arranged in series, while procedures N and R, which can be processed in parallel, are arranged in parallel.

[0058] As shown in Figure 12, the work plan screen 1201 may include a replacement parts list 1203 that displays the estimated number of replacements for each part based on the work plan. Alternatively, based on the condition of soiling and wear of each part obtained by the learning model 63 for determining soiling and wear, it may determine which parts are reusable and which need to be replaced, and display the number of reusable parts and the number of parts that need to be replaced. For example, in the replacement parts list 1802 shown in the work plan screen 1801 of Figure 18, the number of reusable parts and the number that need to be replaced are shown for each part. Such information is useful for preparing parts in advance.

[0059] The proposed parts replacement is not limited to replacement with new parts. If multiple identical parts are used in the device 100, the replacement of the parts' installation locations may be proposed to equalize wear. The proposal to replace parts' installation locations can be made based on the degree of contamination and wear output by the learning model 63 for determining contamination and wear, and if there is a difference in the degree of contamination and wear among identical parts. The proposal to replace parts' installation locations can be made, for example, by a proposal display 1301 as shown in Figure 13.

[0060] [Regarding the processing procedure] Figure 14 is a flowchart showing the process by which the information processing device 200 generates and updates a learning model 44 for determining work instructions.

[0061] In step S101, the learning model generation / update unit 213 acquires usage history information stored in the device state memory unit 221 and inspection information obtained from various sensors 10. In step S102, the learning model generation / update unit 213 acquires work history information stored in the work history information memory unit 222. In step S103, the learning model generation / update unit 213 acquires post-assembly inspection information and repair failure information for refurbished products stored in the device state memory unit 221. In step S104, the learning model generation / update unit 213 links the three pieces of information acquired in steps S101 to S103, uses them as training data to perform machine learning, and generates and updates a learning model 44. The generated and updated learning model 44 is stored in the learning model memory unit 224.

[0062] Figure 15 is a flowchart showing the process by which the information processing device 200 generates and updates a learning model 63 for determining contamination and wear. In step S201, the learning model generation and update unit 213 acquires usage history information stored in the device state storage unit 221 and inspection information obtained from various sensors 10. In step S202, the learning model generation and update unit 213 acquires work history information stored in the work history information storage unit 222. In step S203, the learning model generation and update unit 213 determines the contamination and wear status from the cleaning and replacement work history in the work history information acquired in step S202. Specifically, the learning model generation and update unit 213 can determine the wear status from the location and reason for replacement of parts, and the contamination status from the cleaning work information. Alternatively, the operator may determine the location and type of contamination and wear based on external images, X-ray images, etc., and input this information to the control unit 210. The information acquired in step S203 is called the actual contamination and wear information. In step S204, the learning model generation / update unit 213 generates and updates a learning model 63 for determining contamination and wear by performing machine learning using the device status information acquired in step S201 and the actual contamination and wear information acquired in step S203. The generated and updated learning model 63 is stored in the learning model storage unit 224.

[0063] Figure 16 is a flowchart showing the process by which the information processing device 200 generates and updates a learning model 83 for proposing additional work and procedure manual updates. In step S301, the learning model generation and update unit 213 acquires usage history information stored in the device state storage unit 221 and inspection information obtained from various sensors 10. In step S302, the learning model generation and update unit 213 acquires work history information stored in the work history information storage unit 222. In step S303, the learning model generation and update unit 213 links the two pieces of information acquired in steps S301 to S302, and uses them as training data to perform machine learning and generate and update the learning model 83. The generated and updated learning model 83 is stored in the learning model storage unit 224.

[0064] Figure 17 is a flowchart showing the procedure for processing work instructions by the information processing device 200. First, the worker prepares the device 100 to be remanufactured. Specifically, the worker sets the device 100 in a designated work location and connects it to the work instruction system 1 so that it can communicate with it. After that, the process shown in Figure 17 begins.

[0065] In step S401, the usage history information acquisition unit 211 acquires usage history information from the usage history information storage unit 101 of the device 100 to be remanufactured and stores it in the device state storage unit 221. The method for acquiring usage history information is not particularly limited. In this embodiment, usage history information stored in the non-volatile memory built into the device 100 is read, but for example, if there is a management server that manages the usage history information of the device 100, the necessary data may be acquired from the management server via the network. In step S402, the inspection information acquisition unit 212 acquires inspection information of the device 100 from various sensors 10 and stores it in the device state storage unit 221.

[0066] In step S403, the work instruction / status determination unit 214 obtains the usage history information acquired in step S401 and the inspection information acquired in step S402 from the device status storage unit 221. This information is used as device status information. Subsequently, the work instruction / status determination unit 214 uses the learning model 44 for determining work instructions, the device status information, and the work instruction information stored in the work instruction information storage unit 223 to acquire and output work instruction information. In addition, the additional work / procedure manual update determination unit 216 uses the learning model 83 for proposing updates to additional work / procedure manuals, the device status information, and the work instruction information to acquire and output additional work / procedure manual update proposal information. The work instruction information and the additional work / procedure manual update proposal information may be presented simultaneously or separately.

[0067] In step S404, it is determined whether or not there is a cleaning work instruction among the work instructions output in step S403. This determination is made mechanically (automatically) by adding a cleaning flag to the work instruction information stored in the work instruction information storage unit 223. If there is a cleaning work instruction (Yes in step S404), in step S405, the soiling / wear status determination unit 215 determines the soiling / wear status of the device 100 (i.e., the object to be cleaned). Specifically, the soiling / wear status determination unit 215 uses device status information (usage history information, device status) and a learning model 63 for determining soiling / wear to acquire information on soiled / worn locations and add it to the work instruction. For example, if an external image was acquired during inspection, it is possible to use the image to indicate the soiled / worn locations. If there is no cleaning work instruction (No in step S404), step S405 is skipped and the process proceeds to step S406.

[0068] In step S406, the control unit 210 waits until the worker completes the disassembly and cleaning work in accordance with the work instructions in step S403 and the cleaning instructions in step S405. If a target time for the disassembly and cleaning work is set, the control unit may wait until that target time has elapsed. Alternatively, the completion of the work may be detected by the worker inputting the completion of the work by pressing a button or by sensing the worker's movements and position with sensors.

[0069] In step S407, the inspection information acquisition unit 212 acquires inspection information to be used when determining the completion of the work in step S408. Therefore, the same inspection as in step S402 is performed, and the newly acquired post-work inspection information is stored in the device state storage unit 221.

[0070] In step S408, the work instruction / status determination unit 214 uses the inspection information acquired in step S407 to determine whether the work has been completed according to the work instructions output in step S403. If the same instructions as those already issued are output, it is determined that there was a work omission or that the work was insufficient. If the work is completed (Yes in step S408), the process proceeds to step S410. If it is determined that there is incomplete work (No in step S408), in step S409, the work instruction / status determination unit 214 outputs instructions for the incomplete work again and returns to step S406. By performing work completion determination in this manner, it is possible to automatically check whether the work has been performed appropriately according to the instructions and, if necessary, to automatically instruct additional work, thereby improving work quality and ensuring the quality of remanufactured products.

[0071] In step S410, the learning model generation and update unit 213 updates the learning model 83 using the work history information determined to be completed in step S408, the usage history information acquired in step S401, and the inspection information acquired in step S407. The updated learning model 83 is stored in the learning model storage unit 224. By updating the learning model 83 each time by feeding back the work results in this way, it is possible to optimize additional work instructions and procedure manuals.

[0072] In step S411, the work instruction / status determination unit 214 determines whether all disassembly and cleaning work has been completed. If all work is completed (Yes in step S411), the disassembly and cleaning work is completed and the processing procedure ends. If all work is not completed (No in step S411), the inspection information obtained in step S407 is used to start the next step, and the process is repeated from step S403.

[0073] After disassembly and cleaning are completed, the target device 100 is reassembled as a refurbished product using recycled and replacement parts, and the inspection information performed immediately before shipment is stored in the device state storage unit 221. Subsequently, the inspection information can be linked to the work history by means of work ID or other means, and used to update the learning model 44 for determining work instructions. In this way, by updating the learning model 44 each time using the device state information (mainly inspection information) of the remanufactured target device 100, the accuracy of work instructions can be improved.

[0074] As described above, appropriate work instructions can be given according to the individual state of the target device 100, which can lead to increased work efficiency and more effective use of parts.

[0075] The present disclosure has been described in detail above based on preferred embodiments, but the present disclosure is not limited to those described above and can be modified in various ways without departing from its essence. For example, in the above embodiment, a copier was given as an example of the device 100 to be remanufactured, but a multifunction printer or other office equipment, or other industrial products may be used as the device. Also, in the above embodiment, two types of device status information, usage history information and inspection information, were used for the device status information of the device 100, but work instructions may be generated from only one of the types of information. Also, in the above embodiment, disassembly and cleaning work was performed on the device 100, but only one of the work contents may be performed. If only disassembly work is performed, steps S404 to S405 in Figure 17 may be omitted, and the soiling / wear status determination unit 215 and the learning model 63 for determining soiling / wear may also be unnecessary. Furthermore, the present disclosure can also be understood as a method for manufacturing a remanufactured product, which includes a process of performing remanufacturing work on the device according to the work instructions output from the work instruction system 1.

[0076] This disclosure is not limited to the embodiments described above, and various modifications and alterations are possible without departing from the spirit and scope of this disclosure. Accordingly, the following claims are attached to make the scope of this disclosure public.

[0077] This application claims priority based on Japanese Patent Application No. 2024-224123, filed on December 19, 2024, and all of its contents are incorporated herein by reference.

[0078] 1: Work instruction system 10: Sensor 100: Target device 210: Control unit 220: Memory unit 230: Display unit 44: Learning model for determining work instructions (first learning model) 63: Learning model for determining soiling / wear status (second learning model) 83: Learning model for determining whether additional work / procedure manuals need to be updated (third learning model)

Claims

1. A work instruction system for outputting work instructions for remanufacturing, comprising: a first learning model that has learned by machine learning the correspondence between the state of a device and work instructions for work to be performed on the device in that state; a storage unit that stores device state information indicating the state of a target device to be remanufactured; a control unit that acquires information indicating work instructions for work to be performed on the target device by inputting the device state information of the target device acquired from the storage unit into the first learning model; and a display unit that outputs work instructions for the target device based on the information acquired from the first learning model by the control unit.

2. The work instruction system according to claim 1, wherein the device status information of the target device includes usage history information indicating the usage history of the target device and inspection information indicating the results of inspection of the target device.

3. The work instruction system according to claim 1, wherein the first learning model is machine-trained using information indicating the state of the device sample before remanufacturing, information indicating the work performed on the device sample, and information indicating the state of the device sample after remanufacturing as training data.

4. The work instruction system according to claim 3, wherein the information indicating the state of the device sample after remanufacturing includes at least one of the following: information indicating the results of inspection of the device sample after remanufacturing and information indicating repairs or malfunctions performed after the device sample after remanufacturing has been shipped.

5. The work instruction system according to claim 1, wherein the control unit determines whether the work has been completed in accordance with the work instructions after the work has been performed on the target device according to the outputted work instructions, and if it determines that there are any incomplete tasks, it outputs instructions for the incomplete tasks again.

6. The work instruction system according to claim 5, wherein the control unit inputs information indicating the state of the target device after the work has been performed in accordance with the outputted work instruction to the first learning model, and determines whether the work has been completed in accordance with the work instruction based on the output of the first learning model.

7. The work instruction system according to claim 1, wherein the control unit updates the first learning model using information indicating the state of the target device after remanufacturing.

8. The work instruction system according to claim 1, wherein the storage unit further stores a second learning model that has been machine-learned to determine the correspondence between the state of the device and the location and nature of the contamination and wear of the device in that state, and the control unit obtains information indicating the location and nature of the contamination and wear of the target device by inputting the device state information of the target device into the second learning model.

9. The work instruction system according to claim 8, wherein the display unit displays information indicating the location of the soiling or wear of the target device, along with work instructions for the soiling or wear.

10. The work instruction system according to claim 9, wherein the information indicating the location of soiling or wear of the target device includes an image obtained by imaging the target device.

11. The work instruction system according to claim 8, wherein the control unit, based on information indicating the location and nature of the contamination and wear of the target device obtained from the second learning model, determines that identical parts with different degrees of contamination and wear are used in the target device, and proposes replacing the installation location of the identical parts.

12. The work instruction system according to claim 1, wherein the display unit displays information indicating the overall flow of all tasks to be performed on the target device.

13. The work instruction system according to claim 12, wherein the display unit shows the overall flow of all tasks to be performed on the target device in a flowchart in which the procedure manuals corresponding to each task are arranged in the order of execution.

14. The work instruction system according to claim 13, wherein the flowchart is described in such a way that it can identify which of the above-mentioned tasks can be processed in parallel.

15. The work instruction system according to claim 1, wherein the display unit displays information indicating the reusable parts and their quantities in the remanufacturing of the target device.

16. The work instruction system according to claim 1, wherein the storage unit further stores a third learning model that has been machine-learned to determine the correspondence between the state of the device and any additional work required for the device in that state, and the control unit inputs the state information of the target device into the third learning model to obtain information indicating any additional work that may be required for the work on the target device, and proposes updating the procedure manual corresponding to the work on the target device based on that information.

17. A work instruction method for outputting work instructions for remanufacturing by a computer, comprising the steps of: acquiring information indicating work instructions for work to be performed on a target device by inputting device state information indicating the state of a target device to be remanufactured into a first learning model which has been machine-learned to determine the correspondence between the state of the device and work instructions for work to be performed on the device in that state; and the computer outputting work instructions for the target device based on the information acquired from the first learning model.

18. A program for causing a computer to execute each step of the work instruction method described in claim 17.

19. A method for manufacturing a remanufactured product, comprising the steps of: preparing a target device to be remanufactured; providing information indicating the state of the target device to a work instruction system according to any one of claims 1 to 16; and performing remanufacturing work on the target device in accordance with work instructions output from the work instruction system.

Images