JP7484151B2 - Abnormality diagnosis system and abnormality diagnosis device - Google Patents

Description

The present invention relates to an abnormality diagnosis system and an abnormality diagnosis device.

When a problem occurs with a multifunction device (hereinafter sometimes referred to as an "MFP (Multi Function Printer)") installed at a customer's premises, a service technician visits the customer's premises where the problematic MFP is installed to resolve the issue. The service technician diagnoses the MFP where the problem occurred and takes action based on the diagnosis results. If the action is appropriate, the problem is resolved. The service technician diagnoses the MFP, for example, by referring to error information output by the MFP (for example, a code corresponding to the type of malfunction).

There are various causes of malfunctions, and depending on the malfunction, observing the waveform of the control state may be an effective way to identify the cause of the problem. However, it is fundamentally difficult to bring a measuring device (e.g., a power meter) to the customer's site and observe the waveform by connecting the measuring device to the electronic devices (e.g., sensors, boards, power supplies, etc.) built into the MFP. Therefore, in such cases of malfunctions, service personnel obtain log data from the MFP and analyze the log data to identify the cause. Since log data is made up of a huge amount of numerical information and the know-how for analyzing it is not common, depending on the skill of the analyst, it may take a long time to identify the cause, or it may not be possible to identify the cause at all.

A system has also been developed that identifies the cause of a failure in an image forming device by using a failure diagnosis model that models the causes of the failure of the image forming device (see, for example, Patent Document 1). In the technology described in Patent Document 1, feature amounts of image defects are extracted, and information related to the extracted feature amounts and internal information of the image forming device are input into the failure diagnosis model.

JP 2007-074290 A (paragraphs 0006 to 0008, FIG. 2)

However, the technology described in Patent Document 1 uses the feature quantities of image defects as input, making it difficult to determine the cause of defects that do not appear as image defects. In other words, it is not suitable for determining the cause of defects that require waveform observation as described above.

The present invention was made in consideration of the above-mentioned problems, and the object of the present invention is to provide an abnormality diagnosis system and an abnormality diagnosis device that can broadly investigate the causes of various malfunctions.

The above object of the present invention is achieved by the following means:

(1) An abnormality diagnosis system comprising: an image processing device having a memory unit that stores information related to control as log data; a setting means for inputting work information regarding work performed to address a malfunction in the image processing device; and a machine learning means for determining a cause of the malfunction in the image processing device from the log data and the work information based on machine learning, wherein the machine learning means comprises a first machine learning unit that determines a first cause of the malfunction from the log data based on machine learning, and a second machine learning unit that determines a second cause that is more detailed than the first cause from the first cause and the work information .

(2) An abnormality diagnosis system comprising: an image processing device having a memory unit that stores information related to control as log data; and a machine learning means that determines the cause of a malfunction in the image processing device from the log data based on machine learning, wherein the log data is expressed by one or both of numerical information and text information, the machine learning means performs machine learning using the log data in normal operating states and the log data in abnormal operating states for each operating state of the image processing device, and the operating state is at least one of a warm-up operation, an idling operation, and an operation during printing.

(3) An abnormality diagnosis system comprising: an image processing device having a memory unit that stores information related to control as log data; a machine learning means that determines a cause of a malfunction in the image processing device from the log data based on machine learning; and an input means that inputs a true cause of the malfunction, wherein the machine learning means acquires the true cause and further re-learns.

(4) The abnormality diagnosis system described in (3) above, characterized in that the machine learning means obtains the true cause by inputting information from the person who responded to the malfunction, obtaining information from a data server that manages information about the malfunction, and obtaining log data after the response is completed.

(5) An abnormality diagnosis system comprising: an image processing device having a memory unit that stores information relating to control as log data; and a machine learning means that determines a cause of a malfunction in the image processing device from the log data based on machine learning, wherein the log data is expressed by one or both of numerical information and textual information, and the machine learning means determines the cause of the malfunction from the log data over time or a graph obtained by processing the log data over time.

(6) An abnormality diagnosis device comprising: an image processing device having a memory unit that stores information related to control as log data; a setting means for inputting work information on work performed to address a malfunction in the image processing device; and a machine learning means capable of communicating with one or more of storage means for storing at least one of the log data and the work information, acquiring the log data and the work information of the image processing device in which a malfunction has occurred, and determining the cause of the malfunction in the image processing device from the log data and the work information acquired based on machine learning, wherein the machine learning means comprises a first machine learning unit that determines a first cause of the malfunction from the log data based on machine learning, and a second machine learning unit that determines a second cause that is more detailed than the first cause from the first cause and the work information.

(7) An abnormality diagnosis device comprising: an image processing device having a memory unit that stores information related to control as log data; and a machine learning means that is capable of communicating with an image processing device having a memory unit that stores information related to control as log data or a storage means that accumulates the log data, and that acquires the log data of the image processing device in which a malfunction has occurred and determines the cause of the malfunction in the image processing device from the log data acquired based on machine learning, wherein the log data is expressed by one or both of numerical information and text information, the machine learning means performs machine learning using the log data in normal times and the log data in abnormal times for each operating state of the image processing device, and the operating state is at least one of a warm-up operation, an idling operation, and an operation during printing.

(8) An abnormality diagnosis device comprising: an image processing device having a memory unit that stores information related to control as log data; an input means for inputting the true cause of the malfunction; and a machine learning means that is capable of communicating with one or more of a storage means that stores at least one of the log data and the true cause of the malfunction, and that acquires the log data of the image processing device in which a malfunction has occurred and determines the cause of the malfunction in the image processing device from the log data acquired based on machine learning, wherein the machine learning means acquires the true cause and further re-learns.

(9) The abnormality diagnosis device described in (8 ) above, characterized in that the machine learning means obtains the true cause by inputting information from the person who responded to the malfunction, obtaining information from a data server that manages information about the malfunction, and obtaining log data after the response is completed.

(10) An abnormality diagnosis device comprising: an image processing device having a memory unit that stores information related to control as log data; and a machine learning means that is capable of communicating with the image processing device having a memory unit that stores information related to control as log data, and that acquires the log data of the image processing device in which a malfunction has occurred, and determines the cause of the malfunction in the image processing device from the log data acquired based on machine learning, wherein the log data is expressed by one or both of numerical information and text information, and the machine learning means determines the cause of the malfunction from the log data over time or the log data over time processed and graphed.

The present invention makes it possible to broadly investigate the causes of various defects.

1 is a schematic configuration diagram of an abnormality diagnosis system according to a first embodiment of the present invention; FIG. 13 is an illustration of a learning method in the first machine learning unit. The contents of the log data at each time are shown. The contents of the log data at each time point are converted into waveforms (graphs). 13 is an example of a cause of occurrence report screen. 4 is an example of a flowchart showing a repair process using the abnormality diagnosis system according to the first embodiment of the present invention. FIG. 5 is a schematic configuration diagram of an abnormality diagnosis system according to a second embodiment of the present invention. 13 is an example of a work information input screen. FIG. 13 is an image diagram of a diagnostic method in the second machine learning unit. 13 is an example of a flowchart showing a repair process using the abnormality diagnosis system according to the second embodiment of the present invention.

Below, the embodiment of the present invention will be described in detail with reference to the drawings as appropriate. Each figure is merely a schematic illustration to allow a sufficient understanding of the present invention. Therefore, the present invention is not limited to the illustrated examples. In addition, in each figure, common or similar components are given the same reference numerals, and duplicate explanations thereof will be omitted.

[First embodiment]
Configuration of Abnormality Diagnosis System According to First Embodiment
The configuration of an abnormality diagnosis system 1A according to the first embodiment will be described with reference to Fig. 1. Fig. 1 is a schematic configuration diagram of the abnormality diagnosis system 1A according to the first embodiment.

The abnormality diagnosis system 1A is a system that estimates the cause of a malfunction in a device. There is no particular limitation on the device that is to be diagnosed for malfunctions. Therefore, the abnormality diagnosis system 1A can be used in various situations to diagnose malfunctions in various devices. Devices to be diagnosed are, for example, image processing devices such as printers, copiers, and multifunction peripherals (MFPs). Here, the description assumes that the device is a multifunction peripheral (MFP). Note that a state in which some kind of malfunction has occurred in a "normal" state may be expressed as "abnormal".

The abnormality diagnosis system 1A provides a service to assist the user in repairing the MFP3, and responds to inquiries from the user by estimating the cause of the abnormality in the MFP3. The user in this case is, for example, a serviceman who repairs malfunctions in the MFP3 installed at a customer's site. The serviceman is assigned a predetermined area for which he is responsible, and repairs the MFP3 in that area. Note that a method of dealing with the abnormality may be provided instead of or in addition to the cause.

The abnormality diagnosis system 1A includes an abnormality diagnosis device 2A, one or more MFPs 3, a user terminal 4, and a repair report management DB 5. The abnormality diagnosis device 2A and the MFP 3 can communicate with each other, and the abnormality diagnosis device 2A and the user terminal 4 can communicate with each other. The repair report management DB 5 can be connected to the abnormality diagnosis device 2A and the user terminal 4. The MFP 3 is installed, for example, at a customer's site. The user terminal 4 is distributed to a user in advance, for example. The abnormality diagnosis device 2A and the repair report management DB 5 are installed, for example, in a data center of a person who manages the MFP 3. The abnormality diagnosis device 2A and the repair report management DB 5 may be configured as a cloud system.

The MFP3 shown in FIG. 1 is a device that combines functions such as printing, copying, scanning, and faxing in one device. The MFP3 has a printing unit that prints images on a print medium (e.g., paper). The printing unit forms images based on various data on paper using an electrophotographic imaging process that includes the steps of charging, exposing, developing, transferring, and fixing. The printing unit includes, for example, a plurality of image forming units that form toner images of each color component, an intermediate transfer belt onto which the toner images of each color component are sequentially transferred and which holds the toner images, a secondary transfer device that transfers the toner images on the intermediate transfer belt to paper all at once, and a fixing device that fixes the image that has been secondarily transferred to the paper. The imaging process may be other than the electrophotographic process, and may be other methods such as an impact method, a thermal transfer method, or an inkjet method.

The MFP3 also has a storage unit that stores information related to control as log data (also called a log file). Note that in FIG. 1, the log data is written as "Log", and hereafter the log data may be abbreviated to "Log". The MFP3 has a communication function, and for example, in response to a user operation, transmits log data (which may be only information related to an abnormality) to the abnormality diagnosis device 2A.

The log data may include various information related to the MFP3. Furthermore, the format and structure of the log data are not particularly limited. For example, the log data may represent control information of the MFP3 as numerical information (a sequence of numerical values) or textual information (a sequence of characters). Furthermore, the log data may be configured, for example, with areas divided for each control object (i.e., it may be represented in a categorized manner).

The log data includes a timestamp. The timestamp is information about the date and time. In addition, the log data is preferably configured to associate an operating state with control information so that the operating state can be identified. The operating state here is, for example, a warm-up operation, an idling operation, an operation during printing, etc. This makes it possible to collect log data for each operating state in the machine learning described below, and to perform machine learning on an operating state basis.

The log data includes, for example, paper information, fixing control status, fixing target temperature, fixing detection temperature, heater on status, heater on duty, fixing pressure status, etc. Note that the log data may include at least one of these.

Paper information is information about the paper being printed. Paper information includes, for example, paper size, basis weight, paper type, brand, tray settings, and the like. Note that paper information may include at least one of these. Paper size is information about the size of the paper (A size, B size, etc.). Basis weight is information about the weight of the paper. Paper type is information about the type of paper. Brand is information about the name of the paper. Tray settings is information about the tray from which paper is fed.

The fixing control state is the control state of the fixing device (e.g., heating, cooling, etc.). The fixing target temperature is the target temperature of the fixing device. The fixing target temperature is set, for example, based on the fixing control state. The fixing detection temperature is the detection temperature of the fixing device detected by a sensor. The heater lighting state is information (ON, OFF) indicating the lighting of the heater. The heater lighting duty is the ratio of the time of "High (ON)" in one cycle of "High (ON) + Low (OFF)" in PWM (Pulse Width Modulation) control. Here, PWM control is one of the methods of controlling power, and controls the power output by repeatedly switching between "ON" and "OFF". By changing the heater lighting duty, the actual voltage (real voltage) used for control can be controlled without changing the input voltage. The fixing pressure state is information indicating the state of the pressure applied to the paper during fixing.

The log data also includes, for example, high voltage output values, control information during color registration correction operations, control information during image stabilization correction operations, and control information during paper transport. Note that the log data may include at least one of these.

The high voltage output value is information on the voltage value of the components that make up the printing unit (e.g., drum). The control information during color registration correction operation is parameter information related to color registration. Color registration refers to the overlapping of toner images of each color in the tandem method, and the control information here is information for correcting misalignment such as the position and curvature of the scanning line. The control information during image stabilization correction operation is parameter information related to the stabilization of images (e.g., color reproduction) during continuous printing. The control information during paper transport is information related to the control of the components that make up the transport unit (e.g., rollers and sensors).

The user terminal 4 shown in FIG. 1 is a terminal held by a user, for example a tablet computer. The user terminal 4 has a communication function and is capable of communicating with the abnormality diagnosis device 2A. The user terminal 4 also has an input/output function and is capable of inputting data by the user and outputting data to the user. The user terminal 4 is equipped with, for example, a touch panel display. The user terminal 4 is an example of the "setting means" and "input means" in the claims.

The repair report management DB5 shown in FIG. 1 stores a report created by a user who repaired the MFP3. The report is a document intended to provide information about the repair of the MFP3 to relevant parties (e.g., members of the organization to which the user belongs or a customer), and the data format is not particularly limited. The report may be document data created by word processing software, for example, or image data obtained by scanning handwritten notes on paper. The report may include, for example, the date and time when the repair was performed, the location, the cause of the defect, and information about the repair (e.g., replacement parts, work content, adjustment values). After the repair of the MFP3 is completed, the user creates the report using, for example, the user terminal 4 or a terminal not shown (a terminal other than the user terminal 4).

The abnormality diagnosis device 2A shown in FIG. 1 is a device that estimates the cause of a malfunction (abnormality) of an apparatus. The abnormality diagnosis device 2A mainly includes an abnormality diagnosis DB (Data Base) 11A and a machine learning means 12A. Note that the abnormality diagnosis DB 11A may be located outside the abnormality diagnosis device 2A.

The abnormality diagnosis DB 11A stores various information related to the diagnosis of abnormalities in the MFP 3. The abnormality diagnosis DB 11A stores, for example, information (learning data) for the machine learning means 12A to perform machine learning. The information for the machine learning means 12A to perform machine learning may be, for example, log data of the MFP 3 or processed log data. The abnormality diagnosis DB 11A also stores the true cause of the malfunction. The true cause of the malfunction is input, for example, by the user who repaired the MFP 3. The true cause of the malfunction is used for relearning by the machine learning means 12A.

The machine learning means 12A determines the cause of a malfunction in the MFP 3 from the log data based on machine learning. Hereinafter, the cause estimated by the machine learning means 12A may be referred to as the "estimated cause." The machine learning means 12A has a first machine learning unit 121. A learning method and an abnormality diagnosis method in the first machine learning unit 121 will be described with reference to Figures 2 to 4. Here, a case in which the warm-up of the fixing is not completed will be described as an example.

Figure 2 is an image diagram of the learning method in the first machine learning unit 121. Figure 3 shows the contents of the log data at each time (i.e., the contents of the log data over time). Figure 4 shows the contents of the log data at each time in a waveform (graph). The horizontal axis (X axis) of Figure 4 is time, and the vertical axis (Y axis) is temperature (°C) and ON/OFF status. Figures 3 and 4 show fixing-related information (here, heater illumination status, sensor detected temperature, and sensor target temperature) from power ON to warm-up when fixing warm-up is not complete.

2, the first machine learning unit 121 reads the log data during normal operation and the log data during abnormal operation as supervised learning data, and performs machine learning to construct an abnormality diagnosis model. The answer (correct answer) in the log data during abnormal operation is the true cause of the malfunction. The abnormality diagnosis model is constructed in advance before abnormality diagnosis is performed. It is desirable to prepare a large amount of log data during normal operation and log data during abnormal operation, and the log data may be collected from a plurality of MFPs 3. The first machine learning unit 121 may, for example, input log data (a string of characters or numbers) accompanying time changes as shown in FIG. 3 as learning data to perform learning, or may input log data accompanying time changes as shown in FIG. 4 that has been further converted into a waveform (graph) (image data) as learning data to perform learning. In other words, the method of reading the log data is not particularly limited, and the log data may be read as an image. The first machine learning unit 121 may perform machine learning for each operating state. The operating state is, for example, a warm-up operation, an idling operation, an operation during printing, etc.

Next, an overview of the diagnosis method by the first machine learning unit 121 will be described. If the fixing warm-up is not completed, possible causes include a heater failure or a sensor failure. For example, referring to the distribution shown in FIG. 4, the heater is on during the warm-up state, and the temperature detected by the sensor has hardly changed since the start, so it is highly likely that the sensor is broken. The first machine learning unit 121 learns, for example, such things and estimates the cause of the abnormality. The estimated cause is notified to the user (e.g., a serviceman) via the MFP 3 or the user terminal 4, and the user performs repairs based on the notified estimated cause.

In the first embodiment, when the user (e.g., a serviceman) has completed the troubleshooting, the user inputs the true cause (cause) of the malfunction that occurred this time from the cause report screen 50 (see FIG. 5) displayed on the panel of the MFP3. The cause report screen 50 has a cause input area 51 and a report button 52, and the user inputs the cause of the malfunction in the cause input area 51 and then presses the report button 52 at the bottom right. This causes the MFP3 to send the input cause of the malfunction to the abnormality diagnosis device 2A, where it is stored in the abnormality diagnosis DB 11A. Note that the method of feedback of the cause of the malfunction may be a method other than that using the MFP3 described here. For example, the cause of the malfunction may be sent from the user terminal 4 to the abnormality diagnosis device 2A, or from the repair report management DB 5 to the abnormality diagnosis device 2A.

<<Operation of the Abnormality Diagnosis System According to the First Embodiment>>
A repair process using the abnormality diagnosis system 1A according to the first embodiment will be described with reference to Fig. 6 (or Figs. 1 to 5 as appropriate). Fig. 6 is an example of a flowchart showing a repair process using the abnormality diagnosis system 1A according to the first embodiment.

When a malfunction occurs in the MFP3, a request for repair is conveyed to a service technician (step S11). The service technician visits the customer's site where the malfunctioning MFP3 is installed, and acquires log data from the MFP3 (step S12).

Then, the service technician considers whether or not to analyze the log data with the machine learning means 12A of the abnormality diagnosis device 2A (i.e., whether or not to receive the support provided by the abnormality diagnosis device 2A) (step S13). If the log data is not to be analyzed with the machine learning means 12A ("No" in step S13), the repair process using the abnormality diagnosis system 1A ends. A case in which the log data is not to be analyzed is, for example, a case in which the service technician is able to analyze the log data by himself, in which case the service technician repairs the MFP 3 based on the results of his own analysis. Note that the system may be operated so that the service technician always analyzes the log data with the machine learning means 12A.

On the other hand, if the log data is to be analyzed by the machine learning means 12A ("Yes" in step S13), the serviceman operates the MFP3 to send the log data to the machine learning means 12A (step S14). The machine learning means 12A analyzes the log data by inputting the log data received from the MFP3 to the first machine learning unit 121, and the first machine learning unit 121 outputs an estimated cause (step S15).

Then, the machine learning means 12A of the abnormality diagnosis device 2A feeds back (notifies) the estimated cause to the serviceman (step S16). The machine learning means 12A may feed back the estimated cause to the serviceman by sending the estimated cause to the MFP3, or may feed back the estimated cause to the serviceman by sending the estimated cause to the user terminal 4. The serviceman repairs the MFP3 according to the notified estimated cause (step S17).

After the repair of the MFP 3 is completed, the serviceman operates the MFP 3 or the user terminal 4 to feed back (notify) the true cause (cause of occurrence) of the malfunction that occurred to the machine learning means 12A (step S18). The serviceman inputs the cause of occurrence via, for example, the screen shown in FIG. 5. The machine learning means 12A creates new supervised learning data by associating the log data from which the cause was estimated with the fed back cause of occurrence, and stores the created supervised learning data in the abnormality diagnosis DB 11A. Then, the first machine learning unit 121 of the machine learning means 12A re-learns using the supervised learning data stored in the abnormality diagnosis DB 11A. This improves the accuracy of estimating the cause of the abnormality. The timing of re-learning is not particularly limited and may be any timing. Note that log data of a normal state after repair may be acquired, and re-learning may be performed using the log data after repair.

The abnormality diagnosis system 1A (and the abnormality diagnosis device 2A) according to the first embodiment of the present invention configured as above provides the following advantageous effects.
In other words, the abnormality diagnosis system 1A determines the cause of the abnormality from the log data based on machine learning. Therefore, it is possible to widely investigate the causes of various abnormalities, including those that do not show image defects. In addition, by repeatedly performing re-learning, the accuracy of estimating the cause of the abnormality is improved.

[Second embodiment]
Configuration of the abnormality diagnosis system according to the second embodiment
The configuration of an abnormality diagnosis system 1B according to the second embodiment will be described with reference to Fig. 7. Fig. 7 is a schematic configuration diagram of the abnormality diagnosis system 1B according to the second embodiment. Hereinafter, differences from the abnormality diagnosis system 1A according to the first embodiment (see Fig. 1) will be mainly described.

The abnormality diagnosis system 1B includes an abnormality diagnosis device 2B, one or more MFPs 3, a user terminal 4, and a repair report management DB 5. The communication format in the abnormality diagnosis system 1B is the same as that in the abnormality diagnosis system 1A according to the first embodiment (see FIG. 1).

The abnormality diagnosis device 2B shown in FIG. 7 is a device that estimates the cause of a malfunction (abnormality) of an apparatus. The abnormality diagnosis device 2B mainly includes an abnormality diagnosis DB 11B and a machine learning means 12B. Note that the abnormality diagnosis DB 11B may be disposed outside the abnormality diagnosis device 2B.

The abnormality diagnosis DB 11B stores various information related to the diagnosis of abnormalities in the MFP 3. The abnormality diagnosis DB 11B stores, for example, information (learning data) for the machine learning means 12B to perform machine learning. The information for the machine learning means 12B to perform machine learning is, for example, log data of the MFP 3 (which may be processed log data) and information on work performed on the MFP 3. The abnormality diagnosis DB 11B also stores the true cause of the malfunction. The true cause of the malfunction is, for example, input by the user who repaired the MFP 3. The true cause of the malfunction is used for relearning by the machine learning means 12B.

The machine learning means 12B determines the cause of the malfunction in the MFP 3 based on machine learning from the log data and the work information. In other words, compared to the machine learning means 12A according to the first embodiment, the machine learning means 12B adds work information to the input data. Hereinafter, the cause estimated by the machine learning means 12B may be referred to as the "estimated cause."

As shown in FIG. 7, the machine learning means 12B is composed of a first machine learning unit 121 that learns only log data and a second machine learning unit 122 that learns the work information of the serviceman as additional input data. The first machine learning unit 121 and the second machine learning unit 122, for example, learn for each operating state, and feed back the analysis results to the serviceman. The serviceman performs an analysis based on the feedback results, and improves the accuracy of the machine learning means 12B by feeding back to the machine learning means 12B whether the analysis results match the true cause of the malfunction that occurred. The analysis results may be fed back via the MFP3 or via the user terminal 4. The cause of the malfunction may be fed back via the MFP3, based on the information in the repair report management DB5, or via the user terminal 4.

The method of inputting work information will be described with reference to FIG. 8. FIG. 8 is an example of a work information input screen. The input screen 60 shown in FIG. 8 includes a replacement part selection area 61, a work content input area 62, and an adjustment value input area 63.

When a user (e.g., a serviceman) performing repairs enters work information related to replacement parts during failure analysis, the user presses the replacement parts tab 64 located at the bottom left of the input screen 60. This displays a replacement parts selection area 61 related to the replacement parts, and the user selects the content corresponding to the replacement parts. Replacement parts are parts replaced by the user, such as a photoconductor or a developing unit.

When the user inputs work information related to the work content during defect analysis, the user presses the work content tab 65 located in the center of the bottom of the input screen 60. This displays a work content input area 62 related to the work content, and the user inputs the work content. The work content broadly includes work other than part replacement and parameter adjustment, such as cleaning methods and cleaning locations.

When the user inputs work information related to adjustment values during defect analysis, the user presses the adjustment value tab 66 located at the bottom right of the input screen 60. This displays an adjustment value input area 63 related to the adjustment value, and the user inputs the changed adjustment value. The adjustment value is the value of various parameters, such as the loop amount, resist amount, high pressure output value, and air volume.

The learning method and diagnosis method in the second machine learning unit 122 will be described with reference to FIG. 9. As described in the first embodiment, if the fixing warm-up is not completed, possible causes include heater failure and sensor failure. If the first machine learning unit 121 determines that the presumed cause is sensor failure based on the log data, various other causes are possible, such as the sensor itself being broken, the cable being cut, the circuit board being broken, or the sensor being dirty.

Therefore, in the second embodiment, the user inputs information on the work actually performed (work information) (here, "sensor replacement") and transmits it to the second machine learning unit 122. The second machine learning unit 122 determines a further presumed cause from the presumed cause of the first machine learning unit 121 (here, "sensor failure") and the work information input by the user (here, "sensor replacement"). For example, as shown in FIG. 9, it is assumed that the further presumed cause is determined to be "wire bundle failure". Then, by feeding back the result of the determination to the user, it becomes possible to discover the cause of the occurrence at an early stage. In other words, the second machine learning unit 122 can easily narrow down the cause by acquiring the work information from the user, and can determine a further presumed cause by a process of elimination. As a result, the user does not have to perform all repairs related to the presumed cause of the first machine learning unit 121, and excessive repairs can be suppressed.

<<Operation of the abnormality diagnosis system according to the second embodiment>>
A repair process using the abnormality diagnosis system 1B according to the second embodiment will be described with reference to Fig. 10 (or Figs. 7 to 9 as appropriate). Fig. 10 is an example of a flowchart showing a repair process using the abnormality diagnosis system 1B according to the second embodiment.

Steps S21 and S22 are the same as steps S11 and S12 in the first embodiment. That is, when a malfunction occurs in the MFP3, a request for repair is conveyed to a service technician (step S21). The service technician visits the customer's site where the malfunctioning MFP3 is installed, and acquires log data from the MFP3 (step S22).

Then, the serviceman considers whether or not to analyze the log data with the machine learning means 12B of the abnormality diagnosis device 2B (i.e., whether or not to receive the support provided by the abnormality diagnosis device 2) (step S23). If the log data is not to be analyzed with the machine learning means 12B ("No" in step S23), the repair process using the abnormality diagnosis system 1B ends. A case in which the log data is not to be analyzed is, for example, a case in which the serviceman is able to analyze the log data himself, in which case the serviceman repairs the MFP 3 based on the results of his own analysis. Note that the system may be operated so that the serviceman constantly analyzes the log data with the machine learning means 12B.

On the other hand, if the log data is analyzed by the machine learning means 12B ("Yes" in step S23), the serviceman considers whether or not to input work information, and inputs work information as necessary (step S24). If work information is not to be input ("No" in step S24), the process proceeds to step S25A. On the other hand, if work information is to be input ("Yes" in step S24), the process proceeds to step S25B.

If work information is not to be input ("No" in step S24), the serviceman operates the MFP3 to send log data to the machine learning means 12B (step S25A). The machine learning means 12B inputs the log data received from the MFP3 to the first machine learning unit 121 to analyze the log data, and the first machine learning unit 121 outputs an estimated cause (step S26A). The processing of steps S25A and S26A is similar to the processing of S14 and S15 (see FIG. 6) in the first embodiment.

When inputting work information ("Yes" in step S24), the serviceman operates the MFP3 to send the log data and work information to the machine learning means 12B (step S25B). The machine learning means 12B analyzes the log data and work information received from the MFP3 using the method described in FIG. 9, and outputs an estimated cause (step S26B).

Then, the machine learning means 12B of the abnormality diagnosis device 2B feeds back (notifies) the estimated cause to the serviceman (step S27). The machine learning means 12B may feed back the estimated cause to the serviceman by sending the estimated cause to the MFP3, or may feed back the estimated cause to the serviceman by sending the estimated cause to the user terminal 4. The serviceman repairs the MFP3 according to the notified estimated cause (step S28).

After the repair of the MFP 3 is completed, the serviceman operates the MFP 3 or the user terminal 4 to feed back (notify) the true cause (cause of occurrence) of the malfunction that occurred to the machine learning means 12B (step S29). The serviceman inputs the cause of occurrence via, for example, the screen shown in FIG. 5. The machine learning means 12B creates new supervised learning data by associating the log data and work information from which the cause was estimated with the fed back cause of occurrence, and stores the created supervised learning data in the abnormality diagnosis DB 11B. Then, the machine learning means 12B re-learns using the supervised learning data stored in the abnormality diagnosis DB 11B. This improves the accuracy of estimating the cause of the abnormality. The timing of re-learning is not particularly limited and may be any timing. Note that log data of a normal state after repair may be acquired, and re-learning may be performed using the log data after repair.

The abnormality diagnosis system 1B (and the abnormality diagnosis device 2B) according to the second embodiment described above can also achieve substantially the same effects as those of the first embodiment.
In addition, the abnormality diagnosis system 1B according to the second embodiment can easily narrow down the causes by acquiring work information from the user, and can determine further presumed causes by a process of elimination. As a result, the user does not need to perform all repairs related to the causes presumed by the first machine learning unit 121, and can suppress excessive repairs.

The means and methods for performing various processes in the abnormality diagnosis devices 2A and 2B according to each of the above-mentioned embodiments can be realized by either a dedicated hardware circuit or a programmed computer. The above programs may be provided, for example, by a computer-readable recording medium such as a flexible disk or a CD-ROM, or may be provided online via a network such as the Internet. In this case, the programs recorded on the computer-readable recording medium are usually transferred to and stored in a storage unit such as a hard disk (HDD). The above programs may also be provided as standalone application software, or may be incorporated into the software of the device as one of its functions.

1A, 1B Abnormality diagnosis system 2A, 2B Abnormality diagnosis device 3 MFP (image processing device)
4. User terminal (setting means, input means)
5 Repair report management DB
11A, 11B Abnormality diagnosis DB
12A, 12B Machine learning means 121 First machine learning unit 122 Second machine learning unit

Claims (10)

an image processing device including a storage unit that stores information related to the control as log data;
a setting means for inputting operation information for dealing with a malfunction in the image processing device;
a machine learning unit that determines a cause of a malfunction in the image processing device from the log data and the work information based on machine learning ;
The machine learning means includes:
A first machine learning unit that determines a first cause of the malfunction from the log data based on machine learning;
A second machine learning unit that determines a second cause that is more detailed than the first cause from the first cause and the work information.
An abnormality diagnosis system comprising: an image processing device including a storage unit that stores information related to the control as log data;
a machine learning unit that determines a cause of a malfunction in the image processing device from the log data based on machine learning,
The log data is expressed by one or both of numerical information and character information,
the machine learning means performs machine learning using the log data in a normal state and the log data in an abnormal state for each operating state of the image processing device,
the operating state is at least one of a warm-up operation, an idling operation, and an operation in progress of printing;
An abnormality diagnosis system comprising: an image processing device including a storage unit that stores information related to the control as log data;
a machine learning means for determining a cause of a malfunction in the image processing device from the log data based on machine learning;
An input means for inputting the true cause of the defect ,
The machine learning means acquires the true cause and further re-learns.
An abnormality diagnosis system comprising: The machine learning means obtains the true cause by inputting the person who handled the problem, obtaining information from a data server that manages the problem information, and obtaining log data after the problem is solved.
4. The abnormality diagnosis system according to claim 3 . an image processing device including a storage unit that stores information related to the control as log data;
a machine learning unit that determines a cause of a malfunction in the image processing device from the log data based on machine learning,
The log data is expressed by one or both of numerical information and character information,
The machine learning means determines a cause of the malfunction from log data over time or a graph obtained by processing the log data over time.
An abnormality diagnosis system comprising: an image processing device including a storage unit that stores information related to control as log data ; a setting unit that inputs operation information for dealing with a malfunction in the image processing device; and a storage unit that stores at least one of the log data and the operation information ;
a machine learning unit that acquires the log data and the operation information of the image processing device in which a malfunction has occurred, and determines a cause of the malfunction in the image processing device from the log data and the operation information acquired based on machine learning ;
The machine learning means includes:
A first machine learning unit that determines a first cause of the malfunction from the log data based on machine learning;
A second machine learning unit that determines a second cause that is more detailed than the first cause from the first cause and the work information.
An abnormality diagnosis device comprising: The image processing device is capable of communicating with an image processing device having a storage unit that stores information related to the control as log data or a storage unit that stores the log data,
a machine learning unit that acquires the log data of the image processing device in which a malfunction has occurred, and determines a cause of the malfunction in the image processing device from the log data acquired based on machine learning;
The log data is expressed by one or both of numerical information and character information,
the machine learning means performs machine learning using the log data in a normal state and the log data in an abnormal state for each operating state of the image processing device,
the operating state is at least one of a warm-up operation, an idling operation, and an operation in progress of printing;
An abnormality diagnosis device comprising: an image processing device including a storage unit that stores information related to control as log data , an input unit that inputs the true cause of a malfunction, and a storage unit that stores at least one of the log data and the true cause of the malfunction ,
a machine learning unit that acquires the log data of the image processing device in which a malfunction has occurred, and determines a cause of the malfunction in the image processing device from the log data acquired based on machine learning ;
The machine learning means acquires the true cause and further re-learns.
An abnormality diagnosis device comprising: The machine learning means obtains the true cause by inputting the person who handled the problem, obtaining information from a data server that manages the problem information, and obtaining log data after the problem is solved.
9. The abnormality diagnosis device according to claim 8. The image processing device is capable of communicating with an image processing device having a storage unit that stores information related to the control as log data or a storage unit that stores the log data,
a machine learning unit that acquires the log data of the image processing device in which a malfunction has occurred, and determines a cause of the malfunction in the image processing device from the log data acquired based on machine learning;
The log data is expressed by one or both of numerical information and character information,
The machine learning means determines a cause of the malfunction from log data over time or a graph obtained by processing the log data over time.
An abnormality diagnosis device comprising:

Images