Information processing system, information processing method, and computer program

The information processing system addresses the challenge of varying part scales by linking and duplicating data to create accurate machine learning models for maintenance actions in image forming apparatuses, enhancing prediction accuracy.

JP2024164709AInactive Publication Date: 2024-11-27CANON KK
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
JP2023080385
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2023-05-15
Publication Date
2024-11-27
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing machine learning models for predicting maintenance actions in image forming apparatuses face challenges due to varying scales of replacement parts across regions, leading to decreased learning accuracy when collecting data, as units may be recognized as different labels despite being the same correct label, or the amount of data is insufficient for accurate learning.

Method used

An information processing system that collects learning data by linking error history and parts replacement information, duplicates part information to unit information, and generates second linking data to enrich data sets, using machine learning algorithms like SVM to create accurate models.

Benefits of technology

Enhances learning data accuracy by duplicating data based on correspondence relationships for parts and units, improving prediction accuracy for maintenance actions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2024164709000001_ABST
    Figure 2024164709000001_ABST
Patent Text Reader

Abstract

To provide an information processing system, an information processing method, and a computer program that can collect highly accurate learning data.SOLUTION: An information processing method in an information processing system for collecting learning data includes: error history information acquiring means S552 for acquiring error history information of a target device; part replacement information acquisition means S553 for acquiring part replacement information of the target device; association means S554 for generating first association data in which the error history information and the part replacement information are associated with each other; and duplication means S557 for generating second association data in which information on a prescribed replacement part is replaced with information on another replacement part including the prescribed replacement part and associated with the error history information.SELECTED DRAWING: Figure 5
Need to check novelty before this filing date? Find Prior Art

Description

[Technical field]

[0001] The present invention relates to an information processing system, an information processing method, a computer program, etc. that collect learning data. [Background technology]

[0002] For example, in an image forming apparatus such as a multifunction peripheral, when an error, breakdown, or other abnormality occurs, a maintenance person such as a serviceman is dispatched in response to a report of the abnormality. The maintenance person checks the manual and performs maintenance such as replacing parts.

[0003] Cloud computing has also become increasingly popular in recent years. The main feature of cloud computing is that it uses many computing resources to execute data conversion and processing in a distributed manner, and processes requests from many clients in parallel using distributed parallel processing. By using cloud computing, system developers can easily obtain the necessary computing resources, allowing them to focus on developing system functions.

[0004] One of the elements that has a high affinity with cloud computing is AI (Artificial Intelligence). One of the core technologies that realizes AI is machine learning. Machine learning can create a learning model that extracts the features of the data (characteristics, patterns, trends, etc.) by analyzing large amounts of data (big data) with a learning algorithm.

[0005] Since a large amount of computing resources are required to safely store and analyze such large amounts of data, it is often introduced in cloud computing environments. In the maintenance of the image forming devices mentioned above, a method has been proposed in which data collected from multiple image forming devices is learned and the obtained trained model is used to support maintenance work such as part replacement.

[0006] For example, in Patent Document 1, estimated information is obtained based on device information collected from multiple devices and worker information held by a worker such as a serviceman. The worker information includes part information and work status information obtained from the worker's information terminal. This enables a worker such as a serviceman to efficiently perform replacement work. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] JP 2019-211940 A Summary of the Invention [Problem to be solved by the invention]

[0008] On the other hand, there is a demand for a machine learning model that can estimate the appropriate response to provide information that enables maintenance personnel such as servicemen to take appropriate action in response to errors such as breakdowns. In order to create a machine learning model, a large amount of learning data is required that links errors such as breakdowns with the appropriate responses to the errors (e.g., replacement parts, etc.).

[0009] However, in actual operation, the units of replacement parts may differ. For example, even if a specific part A is replaced in one area, a larger scale unit X that includes part A may be replaced in another area.

[0010] As a result, when trying to collect a large amount of data for learning, even if the labels are actually the same (replacement parts), they may be recognized as different correct labels due to the different scales, which can reduce the learning accuracy.

[0011] On the other hand, if learning is done only using data with the same scale, the amount of data available for learning is reduced, and the learning accuracy may also decrease. In particular, in cases where "in many areas, replacement is done on a part-by-part basis, but in some areas, replacement is done on a unit-by-unit basis," the amount of minor data (data targeting units) tends to be small.

[0012] Therefore, an object of the present invention is to provide an information processing system capable of collecting highly accurate learning data. [Means for solving the problem]

[0013] An information processing system for collecting learning data, comprising: an error history information acquiring means for acquiring error history information of a target device; a part replacement information acquisition means for acquiring part replacement information of the target device; A linking means for generating first linking data linking the error history information and the part replacement information; A duplication means for replacing information of a predetermined replacement part with information of another replacement part including the predetermined replacement part, and generating second linked data linked with the error history information; The present invention is characterized by having the following. Effect of the Invention

[0014] According to the present invention, it is possible to provide an information processing system capable of collecting highly accurate learning data. [Brief description of the drawings]

[0015] [Figure 1] 1 is a diagram illustrating an example of a configuration of an information processing system according to a first embodiment of the present invention. [Diagram 2] 1A is a diagram illustrating an example of a hardware configuration of an image forming apparatus 106 according to a first embodiment of the present invention, and FIG. 1B is a diagram illustrating an example of a hardware configuration of an apparatus information collecting server 102, a replacement part information collecting server 104, and a repair location notifying server 105 according to a first embodiment of the present invention. [Diagram 3] 1A and 1B are diagrams showing examples of screens displayed on an information input terminal 103 according to the first embodiment of the present invention. [Figure 4] FIG. 2 is a functional block diagram showing a configuration example of a repair location notifying server 105 according to the first embodiment of the present invention. [Diagram 5] 4A and 4B are flowcharts illustrating an example of an information processing method according to a first embodiment of the present invention. [Figure 6] 1 is a flowchart showing an example of a machine learning process according to Example 1 of the present invention. [Figure 7] 13 is a flowchart of a learning data collection process limited to device target units according to the second embodiment of the present invention. [Figure 8] 13 is a flowchart of a learning data collection process limited to regional target units according to the third embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Hereinafter, the embodiment of the present invention will be described with reference to the drawings. However, the present invention is not limited to the following embodiment. In each drawing, the same members or elements are given the same reference numerals, and duplicated descriptions are omitted or simplified. EXAMPLES

[0017] 1 is a diagram showing an example of the configuration of an information processing system according to a first embodiment of the present invention, in which a plurality of image forming devices 106 are, for example, a digital multifunction peripheral, a facsimile machine, a laser beam printer, a scanner device, etc. In the following embodiment, data generated to generate a machine learning model is called learning data, and data input to an already generated machine learning model for estimation is called input data.

[0018] The learning data collection including appropriate responses to errors in this embodiment is performed using an apparatus information collection server 102, a replacement parts information collection server 104, a repair location notification server 105, etc., as shown in Fig. 1. A machine learning model is created using the learning data, and a service side that receives an error notification of the image forming apparatus uses the model to make an estimation. The estimation result is displayed on a WEB UI of a portal site that can be accessed by a maintenance staff member.

[0019] An apparatus information collection server 102 constitutes an online service system that provides an information collection system from image forming apparatuses. Error history information from a plurality of image forming apparatuses 106 is collected and stored via the network 101.

[0020] Reference numeral 103 denotes an information input terminal held by a maintenance person, and when the maintenance person performs maintenance work, the details of the maintenance work are input into the information input terminal 103. The information input terminal 103 holds the input details of the maintenance work.

[0021] Moreover, the information input terminal 103 transmits the inputted contents of the maintenance work to the replacement part information collection server 104, the repair part notifying server 105, etc., via the network 101. The information input terminal 103 also receives repair part estimation information transmitted from the repair part notifying server 105 via the network 101, and can display it on the screen of the information input terminal 103.

[0022] Reference numeral 104 denotes a replacement part information collection server that collects information from the information input terminal 103 and provides online services. The server receives replacement part information transmitted by the information input terminal 103 via the network 101, and collects and stores the acquired information.

[0023] Reference numeral 105 denotes a repair location notification server that collects information from the device information collection server 102 and the replacement part information collection server 104 and provides online services, and receives various pieces of information via the network 101 .

[0024] The various information includes feedback information transmitted by the information input terminal 103, error history information held by the device information collecting server 102, and replacement part information held by the replacement part information collecting server 104. The repair point notifying server 105 creates and accumulates learning data based on the various information received. In this manner, the information processing system of this embodiment including the repair point notifying server 105 can collect and accumulate learning data.

[0025] 2A is a diagram showing an example of a hardware configuration of an image forming apparatus 106 according to a first embodiment of the present invention. Each of the hardware components is connected to a system bus 202. Reference numeral 201 denotes a CPU serving as a computer that controls the entire apparatus, and performs overall control of various devices connected to the system bus 202 by executing computer programs stored in a ROM 203 or an external memory 206.

[0026] Reference numeral 203 denotes a ROM serving as a storage means, which stores therein various types of data such as programs such as a basic I / O program, font data used in document processing, and template data.

[0027] Reference numeral 204 denotes a RAM that functions as the main memory, work area, etc. of the CPU 201, and is configured so that the memory capacity can be expanded by an optional RAM connected to an expansion port (not shown). Reference numeral 205 denotes a network interface card (NIC), and data is exchanged with external devices via the interface 205.

[0028] Reference numeral 207 denotes an operation panel that displays a screen and accepts user operation instructions via the screen. The operation panel 207 has a display unit such as a liquid crystal panel and buttons for setting the operation mode of the image forming apparatus, displaying the operation status of the image forming apparatus, and performing operations such as copy designation. Reference numeral 208 denotes an external storage unit that functions as a large-capacity memory.

[0029] Reference numeral 209 denotes a device interface, which is a connection interface with an external device that can be connected via USB, etc. The printer 210 uses a known printing technology, and suitable implementations include, for example, an electrophotographic method (laser beam method), an inkjet method, a dye sublimation method (thermal transfer method), and the like.

[0030] The printer 210 outputs image data converted from print data (PDL language, PDF language, etc.). The scanner 211 uses known image reading technology to optically scan a paper document placed on a transparent top plate and convert it into an image. It also continuously reads multiple paper documents placed on the automatic document feeder (ADF) and converts them into images.

[0031] 2(B) is a diagram showing an example of the hardware configuration of the device information collecting server 102, the replacement part information collecting server 104, and the repair part notifying server 105 according to the first embodiment of the present invention. Each of the hardware components is connected to a system bus 220. The hardware configuration of the information input terminal 103 may also be the same as the configuration shown in FIG. 2(B).

[0032] Reference numeral 221 denotes a CPU serving as a computer for controlling the entire apparatus, and performs overall control of various devices connected to a system bus 220 by executing computer programs stored in a ROM 223 or an external memory 226 .

[0033] Reference numeral 222 denotes a GPU, which is a calculation device specialized for vector calculations such as image processing, machine learning, etc. Reference numeral 223 denotes a ROM, which is a storage means, and stores various data such as a basic I / O program inside.

[0034] Reference numeral 224 denotes a RAM that functions as a main memory, work area, etc. for the CPU 221 and GPU 222, and is configured so that the memory capacity can be expanded by an optional RAM connected to an expansion port (not shown). Reference numeral 225 denotes a network interface card (NIC), and data is exchanged with external devices via the network interface card 225.

[0035] Reference numeral 227 denotes an input / output interface, which can display a screen and receive user operation instructions via a display, keyboard, mouse, smartphone, tablet, or other device. Reference numeral 228 denotes an external storage means that functions as a large-capacity memory. Reference numeral 229 denotes a device interface, which is a connection interface with an external device that can be connected via USB or the like.

[0036] 3(A) and (B) are diagrams showing examples of a screen displayed on the information input terminal 103 according to the first embodiment of the present invention. Reference numeral 301 in Fig. 3(A) is an example of an input screen for information to be transmitted to the replacement part information collection server 104. A maintenance technician inputs replacement part information 311 that has actually been replaced.

[0037] The replacement part information 311 includes the product name, the machine number, the part replacement date, the replacement part information, etc. "Product name" is a name that indicates the product type of the image forming device. "Machine number" is a unique ID assigned to each image forming device. "Part replacement date" is the date when the maintenance person replaced the part.

[0038] "Replacement part information" is the abbreviation of the part that was actually replaced by the maintenance technician. In the figure, parts are displayed as an example, but depending on the country, region, device, etc., a unit larger than a part may be displayed here. Note that here, a unit means, for example, something that is composed of multiple parts and can be replaced on a unit basis.

[0039] 3B shows an example of a portal screen 302 for inputting information to be sent to repair location notification server 105, and is displayed on information input terminal 103. The portal screen includes an error detail information display section 321 including information on the image forming device to be repaired, a recommended repair process content display section 322, and a region selection section 323. Error detail information display section 321 displays the product name, machine number, date of error occurrence, and occurred error code of the image forming device to be repaired.

[0040] "Product name" is the name that indicates the product type of the image forming device to be repaired. "Model number" is a unique ID assigned to each image forming device to be repaired. "Error occurrence date" is the date on which the error occurred. "Occurred error code" is a unique character string code that identifies the type of error.

[0041] The recommended repair process content display unit 322 displays the recommended maintenance process that is highly likely to resolve the event (error) as estimated by the repair point notification server 105. The recommended maintenance process includes processes such as replacement, cleaning, repair, and restarting of specific parts. Based on the estimation result, the recommended maintenance process display also displays the likelihood that the process will resolve the event (error) as shown in FIG. 3(B).

[0042] That is, in the example of the recommended repair process display section 322 shown in Fig. 3(B), the probability that the error will be resolved by replacing part A is 80%, 15% for replacing part B, and 5% for replacing part C. Note that, although parts A to C are displayed as an example in Fig. 3(B), it is also possible to display units that are larger than parts for replacement depending on the country, region, device, etc.

[0043] The area selection section 323 is a screen for inputting the area of ​​use of the user of the information input terminal 103. In addition, the scale of the corresponding recommended replacement parts displayed in the recommended repair process content display section 322 is changed depending on the input information on the area of ​​use.

[0044] Specifically, the recommended maintenance process displayed in the recommended repair process content display section 322 is switched between part replacement and unit replacement. Note that the input of the area information does not have to be done by the area selection section 323, and the area information of the user (maintenance staff) of the logged-in information input terminal 103, the area information of the registered image forming device, or the like may be used.

[0045] Fig. 4 is a functional block diagram showing a configuration example of the repair point notifying server 105 according to the first embodiment of the present invention. Note that some of the functional blocks shown in Fig. 4 are realized by causing a CPU or the like as a computer included in the repair point notifying server 105 to execute a computer program stored in a memory as a storage medium.

[0046] However, some or all of these may be realized by hardware. For the hardware, a dedicated circuit (ASIC) or a processor (reconfigurable processor, DSP) may be used. Furthermore, the functional blocks shown in Fig. 4 do not have to be built into the same housing, and may be configured by separate devices connected to each other via signal paths.

[0047] The computer program of the repair location notification server 105 is read from the RAM 224, the storage device 228, a secondary storage device connected via the device I / F 229, or the like, and executed by the CPU 221 or the GPU 222. Access to the outside of the equipment information collection server 102, the information input terminal 103, the replacement part information collection server 104, and the like is performed via the network interface card 225.

[0048] The repair location notification server 105 has, as data storage sections, a device information storage section 401, a part replacement information storage section 402, an estimation result storage section 403, a feedback storage section 404, and a unit information storage section 405.

[0049] It also has a learning / input data management unit 406, a learning execution unit 407, a machine learning model management unit 408, and an estimation execution unit 409. The device information storage unit 401 stores error information and device information of the image forming device received from the device information collection server 102 via the network 101.

[0050] The part replacement information storage unit 402 stores the replacement part information received from the replacement part information collection server 104 via the network 101. The feedback storage unit 404 receives and stores the feedback information inputted at the information input terminal 103 via the network 101.

[0051] The unit information storage unit 405 holds information about various units that contain (include) parts. Specifically, the unit information storage unit 405 holds information about what parts various units contain (include), what units are replaceable for each model (image forming apparatus), what parts / units are available in each region, and the like.

[0052] The learning / input data management unit 406 creates and stores learning data and input data based on each element information stored in the repair point notification server 105. Each element information includes, for example, the following information.

[0053] Image forming device information stored in the device information storage unit 401 Replacement part information stored in the part replacement information storage unit 402 Feedback information held by the feedback storage unit 404 Unit information held by the unit information storage unit 405

[0054] The learning / input data management unit 406 creates and stores learning data when learning a machine learning model, and creates and stores input data when making an estimation using the machine learning model. The learning execution unit 407 acquires the learning data from the learning / input data management unit 406, executes learning based on a pre-specified machine learning algorithm, and generates a machine learning model.

[0055] The machine learning model is stored in the machine learning model management unit 408. Note that the machine learning model may be recreated by repeatedly executing learning at a predetermined time interval or in accordance with changes in the learning data stored in the learning / input data management unit 406.

[0056] The machine learning model management unit 408 stores the machine learning model created by the learning execution unit 407. Note that the machine learning model used for estimation may be changed (replaced) when triggered by reception of a machine learning model from the learning execution unit 407, a condition determination in the machine learning model management unit 408 (for example, the accuracy rate of a new machine learning model exceeds a certain level), or the like.

[0057] The estimation execution unit 409 executes estimation by acquiring input data from the learning / input data management unit 406 and inputting the input data to a machine learning model stored in the machine learning model management unit 408. The estimation result storage unit 403 stores the result of the estimation executed by the estimation execution unit 409.

[0058] The estimation result storage unit 403 transmits the estimation result to the information input terminal 103 via the network 101. Alternatively, the estimation result may be transmitted when a request is received from the information input terminal 103 via the network 101.

[0059] The information input terminal 103 displays the above estimation result on the portal screen, as shown in 322 of Fig. 3(B). As described above, the estimation result in the example of 322 is that the probability that the error will be resolved by replacing part A is 80%, and similarly, in the case of part B, the probability is 15%, and in the case of part C, the probability is 5%.

[0060] 5(A) and (B) are flowcharts showing an example of an information processing method according to the first embodiment of the present invention, and Fig. 5(A) is an overall flowchart of the repair point notifying server 105 in the first embodiment. Note that the operation of each step in the flowchart in Fig. 5(A) is performed sequentially by a CPU or the like as a computer in the repair point notifying server 105 executing a computer program stored in a memory.

[0061] First, in step S502, the repair location notification server 105 collects learning data in the learning / input data management unit 406. Then, in step S503, the collected learning data is used to create a machine learning model in the learning execution unit 407. Here, step S503 functions as a learning step (learning means) that performs learning using the learning data.

[0062] In step S504, the created machine learning model is stored in the machine learning model management unit 408. In addition, if an error occurs in an image forming device, the repair location notification server 105 receives a notification from the image forming device. Then, the repair location notification server 105 performs estimation in the estimation execution unit 409 using the machine learning model generated in the flow of FIG. 5(A).

[0063] The estimation result is transmitted to the information input terminal 103 and displayed as shown in Fig. 3(B) on a portal site which the maintenance staff can access using the information input terminal 103. The maintenance staff can input the actual replacement part information via a screen 301 displayed on the information input terminal 103 as shown in Fig. 3(A).

[0064] 3(A) receives input of replacement part information, and the input replacement part information is transmitted to replacement part information collection server 104. In addition, the maintenance staff can input feedback information (repair results, etc.) via screen 302 of FIG. 3(B) displayed on information input terminal 103.

[0065] The feedback information may be input via a button or check box (not shown) in the recommended repair process display section 322, or via a pop-up, free input, etc. When the feedback information is input on the screen 302 in FIG. 3(B), the input feedback information is transmitted to the repair location notifying server 105.

[0066] Fig. 5(B) is a flowchart showing a detailed example of the learning data collection step (step S502) of the repair point notification server 105 according to the embodiment 1. Note that the operation of each step of the flowchart in Fig. 5(B) is performed sequentially by a CPU or the like as a computer in the repair point notification server 105 executing a computer program stored in a memory.

[0067] In step S551, learning data collection is started. In step S552, error code history information (error history information) is acquired from the device information collection server 102 and stored in the device information storage unit 401. Here, step S552 functions as an error history information acquisition step (error history information acquisition means) that acquires error history information of the image forming device as the target device.

[0068] Table 1 shows an example of error code history information collected by the device information collection server. In Table 1, "Product" is the name that indicates the product type of the image forming device. "Serial number" is a unique ID given to each image forming device. "Area" is the area where the image forming device is installed. "Error occurrence date" is the date on which the error occurred. "Occurred error code" is a unique character string code used to identify the type of error.

[0069] For example, the first line of Table 1 means that "an error with error code Exxx-yyyy occurred on 2023 / 06 / 01 for product aaa installed in region A, model number DEV00000." [Table 1]

[0070] Next, in step S553, part replacement information is acquired from the replacement part information collection server 104 and stored in the part replacement information storage unit 402. An example of replacement part information collected by the replacement part information collection server is shown in Table 2. Here, step S553 functions as a part replacement information acquisition step (part replacement information acquisition means) that acquires part replacement information of the target device.

[0071] In Table 2, "Product" is the name that indicates the product type of the image forming device, and "Serial Number" is a unique ID given to each image forming device. "Parts Replacement Date" is the date when the maintenance technician replaced the part. "Replacement Part" is the name of the part that the maintenance technician actually replaced.

[0072] For example, the first line of Table 2 means that "Product aaa, unit number DEV00000, replaced part A on 2022 / 02 / 05." Also, the fifth line of Table 2 means that "Product bbb, unit number DEV00002, replaced unit X on 2022 / 02 / 25." [Table 2]

[0073] Next, in step S554, the error code information and the replacement part information are linked based on the feedback information stored in the feedback storage unit 404. Here, step S554 functions as a linking step (linking means) that generates first linking data that links the error history information and the part replacement information.

[0074] An example of the feedback information is shown in Table 3. In Table 3, "Product" is the name indicating the product type of the image forming device, and "Machine Number" is a unique ID assigned to each image forming device. "Parts Replacement Date" is the date on which the maintenance technician replaced the part. "Occurred Error Code" is a unique character string code for identifying the type of the error. "Replaced Part" is the name of the part that was actually replaced by the maintenance technician.

[0075] For example, the first line of Table 3 means that "Product aaa, machine number DEV00000, experienced error code Exxx-yyyy on 2023 / 06 / 05, and part A was replaced in response." [Table 3]

[0076] In step S554, the error code information and the replacement part information are linked based on the feedback information to generate and store the result as shown in Table 4. Note that the data linking the error code information and the replacement part information shown in Table 4 is referred to as the first linking data here. [Table 4]

[0077] Next, in step S555, a loop process is started for each link between error code information and replacement part information in each row of Table 4, which is the output of step S554. Then, in step S556, unit information storage unit 405 determines whether each replacement part shown in each row of Table 4 is an individual part or a unit.

[0078] That is, the unit information storage unit 405 determines whether the value in the column of each "replacement part" shown in Table 4 is a part or a unit, based on the unit / parts correspondence information held in the unit information storage unit 405. Table 5 shows an example of unit / parts correspondence information held in the unit information storage unit 405.

[0079] "Unit name" is a name that represents a unit. "Containing parts" is a list of parts that a unit contains. For example, the first line of Table 5 means "Unit X contains parts A and D." Note that in this embodiment, "contain" has the same meaning as "include" or "contain."

[0080] If the value in the "Replacement Part" column of Table 4 is included in the "Unit Name" column of Table 5, the replacement part is determined to be a unit. If the value in the "Replacement Part" column of Table 4 is not included in the "Unit Name" column of Table 5, the replacement part is determined to be a part. [Table 5]

[0081] If it is determined in step S556 that it is a unit, the process returns to step S555 without executing any processing and proceeds to the data in the next row of Table 4. If it is determined in step S556 that it is a part, in step S557, the unit including the replaced part is also considered to have been replaced and the replaced part information is duplicated.

[0082] Here, step S557 functions as a duplication step (duplication means) that replaces information about a specified replacement part with information about other replacement parts that contain the specified replacement part, and generates second linked data linked to the error history information.

[0083] That is, for a given row in Table 4, if the "replacement part" is determined to be a part in step S556, a unit containing that part is extracted in the "containing parts" column of Table 5. Then, a row in which the "replacement part" column of the linking data between the error code and the replacement part in Table 4 is replaced with the extracted unit is duplicated and added.

[0084] In step S558, if the loop process has not been completed for all rows of Table 4, the process returns to step S556 via step S555 to repeat the above operations. When the above process has been completed for all rows of Table 4, the loop process ends in step S558 and proceeds to step S559 to end the learning data termination process.

[0085] Table 6 shows an example of the linking (duplicated) results between errors and replacement parts when processing has been completed for all rows in Table 4 in step S559. The duplicated and added data has an "*" added to the replacement part item.

[0086] In addition, in this embodiment, data in which information about a specific replacement part (e.g., part A) is replaced with information about another replacement part (e.g., unit X) that contains the specific replacement part, and linked to error history information, as in Table 6, is referred to as second linked data. [Table 6]

[0087] In this embodiment, in steps S554 to S558, after performing linking, it was determined whether the replacement part was a unit or a part, and the data was copied. That is, information on a specific replacement part in the first linking data was copied by replacing it with information on another replacement part, and a second linking data linked with the error history information was generated.

[0088] However, this process may be reversed, and data may be copied by first judging whether it is a unit or a part, and then the linking may be performed. That is, after replacing the information of a given replacement part with information of another replacement part, second linking data may be generated that links the information of the other replacement part with the error history information.

[0089] Next, Fig. 6 is a flowchart showing an example of machine learning processing according to the first embodiment of the present invention, and shows one example of a machine learning model creation step (step S503) of the repair point notifying server 105. Note that the operation of each step of the flowchart in Fig. 6 is performed sequentially by a CPU or the like as a computer in the repair point notifying server 105 executing a computer program stored in a memory.

[0090] In step S601, creation of a machine learning model is started, and in step S602, the repair point notification server 105 combines the data in Table 6 with the image forming device information stored in the device information storage unit 401 to obtain collected data. At this time, the data in Table 6 is divided into rows for parts and rows for units, and each is obtained as collected data. That is, learning is performed using the first linked data and the second linked data as collected data, respectively.

[0091] After step S603, a learning process is performed using collected data related to the unit. An example of the collected data is shown in Table 7. In Table 7, "Product" is a name indicating the product type of the image forming device. "Operation information 1" and "Operation information 2" are information indicating the usage status of the image forming device, such as the number of sheets of paper printed on the date the data was obtained.

[0092] The "Exxx-yyyy" and "Exxx-zzzz" columns indicate the respective error codes, with a value of 1 if the respective error has occurred and a value of 0 if it has not. The "Unit X" and "Unit Y" columns indicate a value of 1 if the unit contains a replaceable part and a value of 0 if it does not.

[0093] For example, the first line of Table 9 indicates that "for product aaa installed in region A, machine number DEV00000, unit X was replaced for error code Exxx-yyyy that occurred on 2023 / 06 / 01, and the value of operation information 1 is 1 and the value of operation information 2 is 7."

[0094] [Table 7]

[0095] In step S603, the repair location notification server 105 divides the collected data prepared in step S602, such as that in Table 7, into input data and correct answer data. In this embodiment, the input data is the values ​​in the columns of operation information 1, operation information 2, and error code, and the correct answer data is the values ​​in the columns of unit X and unit Y.

[0096] The input data may include values ​​such as time information (such as the time when copying was performed), user attributes (such as department, position, age), and image forming device attributes (such as model, installation location, copying speed).

[0097] Next, in step S604, the repair location notification server 105 determines whether the setting items (unit X, unit Y) of the correct answer data generated in step S603 have been processed one by one for all setting items. If Yes, the learning model creation flow in FIG. 6 ends. If No, proceed to step S605.

[0098] In this embodiment, the correct answer data is processed one setting item at a time, but multiple setting items may be learned together. However, if multiple setting items of the correct answer data are learned together, a large amount of data is required to perform highly accurate estimation. For example, it is difficult to collect a large amount of learning data for each machine number, and the amount of learning data is small, which may result in poor estimation accuracy.

[0099] In step S605, the repair portion notifying server 105 performs learning using a Support Vector Machine (SVM) for one setting item of the input data and the correct answer data generated in step S603.

[0100] SVM is a well-known machine learning algorithm, and in this embodiment, a nonlinear soft margin SVM using RBF (Radial Basis Function Kernel) is used. In addition, when learning, the learning results are evaluated by cross-validation, which randomly divides the learning data into analysis data and validation data.

[0101] The hyperparameters of SVM using RBF include the cost parameter (C) and gamma (γ), which affect the estimation performance. Therefore, a grid search is performed to comprehensively try multiple values ​​for each, and the one with the highest estimation performance is used as the learning result.

[0102] Any algorithm may be used as the machine learning algorithm, such as perceptron, logistic regression, neural network, k-nearest neighbor method, naive Bayes, convolutional neural network, decision tree, random forest, linear regression, polynomial regression, Lasso regression, Ridge regression, etc.

[0103] In addition, there are various types of SVM, such as hard margin SVM and polynomial kernel SVM, and any of them may be used. In addition, the hyperparameters of the machine learning algorithm may be made different depending on the machine learning algorithm. In addition, as a method for evaluating the learning results, alternative estimation methods and test sample methods may be used in addition to the cross-validation method.

[0104] Furthermore, as a method for optimizing hyperparameters of a machine learning algorithm, any of grid search, random search, Latin hypercube sampling, Bayesian optimization, etc., may be used. In this embodiment, the machine learning algorithm, the method for evaluating the learning result, the method for optimizing hyperparameters of the machine learning algorithm, etc. may be appropriately changed or combined.

[0105] In step S606, the repair location notification server 105 saves the trained model trained in step S605 in a file, and registers it in the machine learning model management unit 408. The file in which the trained model is saved includes the type of learning algorithm (nonlinear soft margin SVM using RBF, etc.), the values ​​of hyperparameters of the learning algorithm (cost parameter (C) and gamma (γ)), ​​etc.

[0106] After saving the trained model in step S606, the process returns to step S604 to determine whether correct data has been processed for all items. If the answer is No, steps S605 and S606 are repeated. If the answer is Yes, the process proceeds to step S607, where the flow in FIG. 6 is ended. EXAMPLES

[0107] According to the first embodiment, for part replacement data with different scales, such as individual parts and units, data for learning can be enriched by duplicating the data based on the correspondence. However, there are cases where the units handled differ depending on the product model of the device. If data is duplicated and learned including units that are not actually handled by the product model, data with patterns that are not possible in reality may also be learned, and the learning accuracy may decrease.

[0108] Therefore, in the second embodiment, data duplication is performed only for units that include replaced parts and are handled by the product model. This makes it possible to duplicate data while eliminating patterns that are not actually possible, thereby improving the learning accuracy.

[0109] The system configuration diagram, hardware configuration diagram, input terminal screen example diagram, and software configuration diagram of the second embodiment are similar to those of the first embodiment in Fig. 1 to Fig. 4, respectively, and therefore the explanations thereof will be omitted. Also, the overall flowchart of the repair point notification server 105 is similar to Fig. 5(A) of the first embodiment, and therefore the explanations thereof will be omitted.

[0110] Fig. 7 is a flowchart of a learning data collection process limited to device target units according to the second embodiment of the present invention, and shows details of the learning data collection step (step S502) of the repair point notifying server 105 in the second embodiment. Note that the operation of each step in the flowchart in Fig. 7 is performed sequentially by a CPU or the like as a computer in the repair point notifying server 105 executing a computer program stored in a memory.

[0111] The process contents of steps S551 to S556, S558, and S559 are the same as those shown in Fig. 5B, and therefore the description will be omitted. If it is determined to be a unit in step S556, the process of steps S701 to S703 is not executed, and a loop process is performed to proceed to the next row of data.

[0112] If it is determined in step S556 that the part is a part, in step S701, the device (product model) that performed the part replacement process is identified from the "product" column of the table showing the link between errors and replacement parts shown in Table 4.

[0113] Next, in step S702, the unit including the replaced part and which is a target unit for the relevant product model is identified by referring to the unit / parts correspondence information (including the handled model information) as shown in Table 8. Here, step S702 functions as a model identification step (model identification means) for identifying whether the target device in which the specified part has been replaced is a model that can use other replacement parts.

[0114] In Table 8, "Unit Name" is the name of the unit, and "Containing Parts" is a list of the parts that the unit contains. "Handling Models" is a list of the product models that handle the unit. For example, the first line of Table 8 means that "Unit X contains Parts A and D, and is handled by product models bbb and ccc."

[0115] [Table 8]

[0116] In step S703, if the unit includes the replaced part and is a unit that is applicable to the product model, the unit is also treated as having been replaced and the replaced part information is copied.

[0117] Also, for the unit identified in step S702, the linking data between the error and the replacement part is duplicated, and the replacement part string is replaced with the identified unit. That is, in step S703, when the target device identified in step S702 as the model identification step is a model that can use other replacement parts, second linking data is generated.

[0118] Furthermore, the processes in step S556 and steps S701 to S703 are executed for all rows by the loop processes in steps S555 and S558, and then the process proceeds to step S559, where the learning data collection flow ends.

[0119] Proceed to step S559 and end learning data collection. An example of linking errors and replacement parts (including duplicated models) is shown in Table 9. Duplicated data has an "*" added to the "replacement part" field.

[0120] Comparing with Table 6 in Example 1, it can be seen that the replacement information of part A that occurred in DEV00000 in the first row has not been replicated as unit replacement information. This is because product model aaa is not included in the target models of unit X in Table 8.

[0121] [Table 9]

[0122] In this embodiment, in steps S554 to S703, after performing linking, it is determined whether the replacement part is a unit or a part, and data is copied. However, the order of these processes may be reversed, and it may be determined first whether it is a unit or a part, the data may be copied, and then linking may be performed. In addition, the sequence of creating a machine learning model in the second embodiment may be the same as the sequence shown in FIG. 6 in the first embodiment, so a description thereof will be omitted.

[0123] According to this embodiment, when duplicating data linking errors and replacement parts, by duplicating only units that include the part and are handled by the product model, it is possible to amplify learning data by limiting it to unit replacements that may actually occur. This makes it possible to improve the learning accuracy of parts with different scales when predicting replacement parts for errors. EXAMPLES

[0124] In the first embodiment, it is assumed that data for learning is enriched by duplicating data based on the correspondence between parts and units, which are part replacement data with different scales. However, there are cases where the units handled differ depending on the region. Even in such cases, some parts may be used in multiple units.

[0125] Therefore, if data is replicated for all units that contain a certain part, it is possible that units that are not actually handled in that region will also be trained. As a result, data with patterns that are not possible in reality may be trained, which may result in a decrease in training accuracy.

[0126] Therefore, in the third embodiment, data duplication is performed only for units that contain replaced parts and are handled in the relevant region. This makes it possible to duplicate data while eliminating patterns that are not actually possible, thereby improving the learning accuracy.

[0127] The system configuration diagram, hardware configuration diagram, input terminal screen example diagram, and software configuration diagram of the third embodiment are similar to those of the first embodiment in Fig. 1 to Fig. 4, respectively, and therefore the explanations thereof will be omitted. Also, the overall flowchart of the repair point notification server 105 is similar to Fig. 5(A) of the first embodiment, and therefore the explanations thereof will be omitted.

[0128] Fig. 8 is a flowchart of a learning data collection process limited to a regional target unit according to the third embodiment of the present invention, and shows details of the learning data collection step (step S502) of the repair point notification server 105 in the third embodiment. Note that the operation of each step in the flowchart in Fig. 8 is performed sequentially by a CPU or the like as a computer in the repair point notification server 105 executing a computer program stored in a memory.

[0129] The process contents of steps S551 to S556, S558, and S559 are the same as those shown in Fig. 5B, and therefore the description will be omitted. If it is determined to be a unit in step S556, the process of steps S801 to S803 is not executed, and the loop process proceeds to the next row of data.

[0130] If it is determined in step S556 that the part is a part, in step S801, the region where the part replacement process was performed is identified from the "region" column of the table showing the link between errors and replacement parts shown in Table 4.

[0131] Next, in step S802, the unit including the replaced part and available in the device installation area is identified by referring to the unit / parts correspondence information (including handling area information) as shown in Table 10. Here, step S802 functions as an area identification step (area identification means) that identifies the installation area of ​​the target device in which the specified part has been replaced.

[0132] In Table 10, "Unit Name" is the name that represents the unit, and "Containing Parts" is a list of the parts that the unit contains. "Handling Area" is a list of the areas that handle the unit. For example, the first line of Table 10 means that "Unit X contains Parts A and D, and is handled in Area A, Area B, and Area C."

[0133] [Table 10]

[0134] In step S803, if the unit includes the replaced part and is available in the relevant region, the unit is also treated as having been replaced and the replaced part information is duplicated.

[0135] For the unit identified in step S802, the linking data between the error and the replacement part is duplicated, and the replacement part string is replaced with the identified unit. That is, in step S803, when the installation area identified in step S802 as the area identification step is an area where other replacement parts can be handled, second linking data is generated.

[0136] Furthermore, the processes in step S556 and steps S801 to S803 are executed for all rows by the loop processes in steps S555 and S558, and then the process proceeds to step S559, where the learning data collection flow ends.

[0137] Proceed to step S559, and an example of linking errors and replacement parts (including duplicated and handling areas) at the time when learning data collection is completed is shown in Table 11. Duplicated data has "*" added to the replacement part item.

[0138] Comparing with Table 6 in Example 1, it can be seen in the third and fourth lines of Table 11 that the replacement information of parts B and C that occurred in DEV00001 has not been replicated as unit replacement information. This is because region B is not included in the target models of unit Y in Table 10.

[0139] [Table 11]

[0140] In this embodiment, in steps S554 to S803, after performing linking, it is determined whether the replacement part is a unit or a part, and data is copied. However, the order of these processes may be reversed, and it may be determined first whether it is a unit or a part, the data may be copied, and then linking may be performed. In addition, the sequence of creating a machine learning model in the third embodiment may be the same as the sequence shown in FIG. 6 in the first embodiment, so a description thereof will be omitted.

[0141] According to this embodiment, when duplicating data linking errors and replacement parts, only units that include the part and are handled in the relevant region are duplicated, making it possible to amplify learning data limited to unit replacements that may actually occur. This makes it possible to improve the learning accuracy of parts with different scales when predicting replacement parts for errors.

[0142] The present invention has been described in detail above based on its preferred embodiments, but the present invention is not limited to the above embodiments. Various modifications are possible based on the spirit of the present invention, and these modifications are not excluded from the scope of the present invention.

[0143] For example, in Tables 6, 9, and 11 of the above embodiments, the first linking data and the second linking data are in different rows, but the first linking data and the second linking data may be included in one row, for example, by providing a column for parts and a column for units in the same row.

[0144] Also, although an image forming apparatus has been described as an example of a device to be repaired (target device), the target device may be any device that requires part replacement, such as an automobile, a ship, an aircraft, a plant, etc. The present invention includes the following combinations.

[0145] (Configuration 1) An information processing system for collecting learning data, comprising: an error history information acquisition means for acquiring error history information of a target device; a part replacement information acquisition means for acquiring part replacement information of the target device; a linking means for generating first linking data linking the error history information with the part replacement information; and a duplication means for replacing information of a specified replacement part with information of another replacement part that contains the specified replacement part, and generating second linking data linked to the error history information.

[0146] (Configuration 2) The information processing system according to configuration 1, further comprising a learning means for performing learning using the first linked data and the second linked data.

[0147] (Configuration 3) The information processing system according to configuration 1 or 2, characterized in that the duplication means duplicates the information of the specified replacement part in the first linking data by replacing it with information of the other replacement part, thereby generating the second linking data linked to the error history information.

[0148] (Configuration 4) An information processing system according to any one of configurations 1 to 3, characterized in that the duplication means replaces the information of the specified replacement part with information of the other replacement part, and then generates the second linking data linking the information of the other replacement part with the error history information.

[0149] (Configuration 5) An information processing system according to any one of configurations 1 to 4, characterized in that the target device on which the specified part has been replaced has a model identification means for identifying whether or not it is a model that can use the other replacement part, and the duplication means generates the second linking data when the target device identified by the model identification means is a model that can use the other replacement part.

[0150] (Configuration 6) An information processing system as described in any one of configurations 1 to 5, further comprising an area identification means for identifying an installation area of ​​the target device in which the specified part has been replaced, and the duplication means generates the second linking data when the installation area identified by the area identification means is an area in which the other replacement part can be handled.

[0151] (Method) An information processing method for collecting learning data, comprising: an error history information acquisition step for acquiring error history information of a target device; a part replacement information acquisition step for acquiring part replacement information of the target device; a linking step for generating first linking data linking the error history information with the part replacement information; and a duplication step for replacing information of a specified replacement part with information of another replacement part that contains the specified replacement part, to generate second linking data linked to the error history information.

[0152] (Program) A computer program for controlling each means of the information processing system according to any one of configurations 1 to 6 by a computer.

[0153] In order to realize a part or all of the control in the above-mentioned embodiment, a computer program for realizing the functions of the above-mentioned embodiment may be supplied to an information processing system or the like via a network or various storage media. Then, a computer (or a CPU, MPU, or the like) in the information processing system or the like may read and execute the program. In this case, the program and the storage medium storing the program constitute the present invention. [Explanation of symbols]

[0154] 102: Equipment information collection server 103: Information input terminal 104: Replacement parts information collection server 105: Repair location notification server 106: Image forming device

Claims

1. An information processing system for collecting learning data, comprising: an error history information acquiring means for acquiring error history information of a target device; a part replacement information acquisition means for acquiring part replacement information of the target device; a linking means for generating first linking data linking the error history information with the part replacement information; a replicating means for replacing information on a predetermined replacement part with information on another replacement part that contains the predetermined replacement part, and generating second linked data linked to the error history information; An information processing system comprising:

2. 2. The information processing system according to claim 1, further comprising: a learning unit that performs learning using the first linked data and the second linked data.

3. The information processing system according to claim 1, characterized in that the duplication means duplicates the information of the specified replacement part in the first linking data by replacing it with information of the other replacement part, thereby generating the second linking data linked to the error history information.

4. The information processing system according to claim 1, characterized in that the duplication means replaces the information of the specified replacement part with the information of the other replacement part, and then generates the second linking data that links the information of the other replacement part with the error history information.

5. a model identification means for identifying whether the target device in which the predetermined part has been replaced is a model that can use the other replacement part, 2. The information processing system according to claim 1, wherein the replicating means generates the second linking data when the target device identified by the model identifying means is a model that can use the other replacement part.

6. a region identification means for identifying an installation region of the target device in which the specified part has been replaced; The information processing system according to claim 1, characterized in that the duplication means generates the second linking data when the installation area identified by the area identification means is an area in which the other replacement parts can be handled.

7. An information processing method for collecting learning data, comprising the steps of: an error history information acquisition step of acquiring error history information of a target device; a part replacement information acquisition step of acquiring part replacement information of the target device; a linking step of generating first linking data linking the error history information and the part replacement information; a duplication step of replacing information of a predetermined replacement part with information of another replacement part including the predetermined replacement part to generate second linked data linked to the error history information; 13. An information processing method comprising:

8. A computer program for controlling each unit of the information processing system according to any one of claims 1 to 6 by a computer.

Citation Information

Patent Citations

  • Maintenance system and maintenance method

    JP2019211940A