JP4600237B2 - Failure diagnosis system, image forming apparatus, and failure diagnosis method - Google Patents

Description

  The present invention relates to failure diagnosis in an image forming apparatus such as a copying machine or a printer, in particular, a failure diagnosis system capable of detecting an abnormality or failure of an electronic circuit implemented on a printed wiring board (PWBA), an image forming apparatus, and a failure. It relates to a diagnostic method.

  In recent years, as electronic devices such as personal computers and copiers have been improved in performance and functions, analog and digital electronic circuits for various purposes have been increasingly formed in the form of printed wiring boards (PWBA). Has been stored in. In addition, many electronic circuit boards that are highly reliable and capable of high-speed and high-precision operation are installed as a means for controlling the operation of everything from automobiles, aviation, robots, semiconductor design equipment, and other industrial equipment. ing. In order to realize a series of functions, these electronic circuit boards are connected through cables in various forms, thereby realizing desired specifications. The environment in which such a board-mounted device is used is usually in an office or a house, but may be used in other harsh environments. Over. Under such circumstances, abnormalities and failures in various states that are difficult to detect occur, and a great deal of labor is required to repair them. Even when used in a normal use environment, an abnormality or failure occurs in the electronic circuit board itself, and the frequency is not necessarily low, and the detection location cannot often be specified.

  As described above, when an abnormality occurs in the electronic circuit board, an immediate response is necessary from the aspects of safety and cost, but as a response when an abnormality occurs in a copying machine, a printer, etc. in the market, With reference to past cases, probabilistic analysis of results superimposed on failure phenomena is performed. As such a technique, for example, proposals such as Patent Document 1 and Patent Document 2 have been proposed. In Patent Document 1, a current failure location candidate is estimated as a probability distribution from a failure cause frequency distribution using past cases as a database. Moreover, in patent document 2, it has performed to the input of the treatment etc. which were given to the database in addition to the cause of failure. In this way, not only functionally following the cause of failure, but also by referring to past cases, failure diagnosis including the failure tendency specific to the device is performed, and the case is databased as new failure data. It is done to register and use.

JP-A-5-35484 JP-A-5-61877

  However, the following problems existed in the examples disclosed in Patent Document 1 and Patent Document 2. In other words, it just reflects the measures taken for the cause of the failure as a case, or because the use of internal information is scarce, it is not always possible to construct case data for accurate diagnosis. In addition, if it is attempted to improve the diagnostic accuracy, it is necessary to carry out on-site countermeasures and the accompanying check work very finely, so that the work becomes complicated.

  The present invention has been made in consideration of the above points, and improves the accuracy of fault diagnosis of a circuit such as PWBA mounted on a device as compared with a fault diagnosis system that refers only to examples. An object of the present invention is to provide a fault diagnosis system capable of

In order to achieve the above object, a failure diagnosis system of the present invention is a failure diagnosis system for an image forming apparatus, which connects a cause of a failure of the image forming apparatus and input information having a causal relationship with the cause of the failure, A failure diagnosis unit that performs inference processing based on a failure diagnosis model that probabilistically extracts abnormality candidates, a log acquisition unit that acquires a log created by describing state information of the image forming apparatus, and the log acquisition unit A log analysis unit that analyzes the acquired log and creates analyzed data, and the failure diagnosis unit is assigned a failure probability value (failure probability value) to the analyzed data in the log analysis unit Inference processing is performed as input information of the fault diagnosis model . Here, the failure diagnosis system may employ a mechanism for performing automatic failure diagnosis using a rule-based system (rule type system, for example, a Bayesian network). Such a failure diagnosis system can construct a failure diagnosis model based on past failure cases that cause an abnormal operation of the apparatus, and can use this failure diagnosis model to obtain abnormality candidates stochastically. In the present invention, further input information to the fault diagnosis model is given in order to improve the accuracy of the fault diagnosis by such a fault diagnosis model. Further input information is information obtained by analyzing and extracting logs exchanged between control systems. The image forming apparatus includes control systems such as ESS (Electric Sub System), IIT (Image Input Terminal), and IOT (Image Output Terminal), and state information of the image forming apparatus is described between these control systems. Logs created may be exchanged. In the present invention, information that can be used for fault diagnosis is extracted from such logs that have been created conventionally and reflected in the fault diagnosis model.

  With such a configuration, it is possible to grasp the state of an electronic circuit incorporated in the image forming apparatus, and to detect an abnormality or failure of an electronic circuit such as PWBA, which has been difficult to grasp by visual observation or the like. Can do.

  Note that log acquisition, log analysis, and the like according to the present invention can be performed by driving software incorporated in elements constituting the image forming apparatus.

In such a fault diagnosis system, and the fault diagnosis unit, and the log acquisition unit, the log analysis unit, the disposed outside of the image forming apparatus may be configured to perform fault diagnosis over the network . With such a configuration, failure diagnosis can be performed remotely.

Further, in such a failure diagnosis system, the log analysis unit extracts a predetermined keyword from the created log to generate the analyzed data, and the failure diagnosis unit is based on the presence or absence of the keyword. It is possible to change the value of the failure probability when inputting the analyzed data as the input information . Further, in this configuration, the log analysis unit creates analyzed data including data related to the description position of the extracted keyword on the log, and the failure diagnosis unit includes the description position of the keyword. The failure probability value when inputting the analyzed data as the input information can be changed based on the above . The log may be created in a text format, for example. Information having a causal relationship with a cause of failure by performing filtering for extracting a keyword from the log created in such a text format. And can be extracted. Filtering can be performed by a conventionally known procedure such as using software. Here, if the analyzed data is created in consideration of the information (description position) where the keyword extracted by filtering is described on the log, more accurate fault diagnosis can be performed.

It should be noted that the image forming apparatus of the present invention can be configured by installing the failure diagnosis system as described above in various image forming apparatuses .

Further, the failure diagnosis method of the present invention is a failure diagnosis method for an image forming apparatus, a process of acquiring a log created by describing status information of the image forming apparatus, and analyzing the log to extracting the keywords, and comprising: the step of creating the parsed data, and performing inference process according to the fault diagnosis model as input information for fault diagnosis model to which the parsed data a failure probability value is assigned, the To do. Such a failure diagnosis method can be implemented, for example, by operating the failure diagnosis system of the present invention.

  According to the present invention, a failure diagnosis model that connects failure causes of an image forming apparatus and input information having a cause-and-effect relationship with the failure causes and extracts failure candidates stochastically leads to failure causes included in the log. By adding information, new abnormality information can be extracted, an abnormality or failure of an electronic circuit can be detected, and the accuracy of failure cause identification can be improved.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the drawings.

  FIG. 1 shows a network to which a failure diagnosis system of the present invention is applied and devices connected thereto. A diagnosis is performed by a desktop personal computer 11 connected remotely or a mobile personal computer 12 connected on-site, or a diagnosis of the image forming apparatus 13 via a network 14 such as a LAN. These personal computers 11 and mobile personal computers 12 fulfill the functions of the failure diagnosis unit, log acquisition unit, and log analysis unit of the present invention, and software that can realize these functions is stored. Further, the image forming apparatus 13 creates a log describing its own state information.

  FIG. 2 shows a configuration example of the PWBA that is a circuit that controls the image forming apparatus 13. Each block is actually composed of a plurality of PWBAs. That is, the ESS 21 includes a plurality of ESS PWBAs and their peripheral elements, the IIT 22 includes a plurality of IIT PWBAs and their peripheral elements, the IOT 23 includes a plurality of IOT PWBAs and their peripheral elements, and the FAX 24 includes a plurality of FAX PWBAs and their peripheral elements. Each of the MCU 25 and the HDD 26 includes peripheral elements. The image forming apparatus 13 further includes RTC 271, NV 272, RAM 273, ROM 274, cable 212 for connecting ESS 21 and IIT 22, cable 213 for connecting ESS 21 and IOT 23, cable 214 for connecting ESS 21 and FAX 24, A cable 215 for connecting the ESS 21 and the MCU 25 and a cable 216 for connecting the ESS 21 and the HDD 26 are provided. The ESS 21 is connected to a PSW (Portable Service Workstation). Such an image forming apparatus 13 is connected to a network capable of external communication.

  FIG. 3 shows the concept of the nodes to be created and the values of the probability table based on the log information created in the image forming apparatus 13 and acquired by the personal computer 11 having the functions of the log acquisition unit and log analysis unit. Fault diagnosis is performed by applying to a node of a Bayesian network that is considered as one of models probabilistically diagnosed based on information (analyzed data) obtained in this way. Probability items related to anomalies that are clarified when a log is acquired and analyzed for a Bayesian network model built in advance are prepared in the node, and the log is actually acquired and analyzed, that is, Filtering As a result, when there is a description that is determined to be an abnormality related to the node, basically, the probability of NG at the time of log acquisition is set to 100% or a value corresponding thereto. In other words, when such analyzed data is obtained, it is assumed that there is a high probability that an abnormality has occurred in the location, and a value corresponding to the probability is input. The log here assumes that the result of checking the internal state of the function related to PWBA when the image forming apparatus 13 is started is output in the form of a text file or the like. Further, the relationship between the presence / absence of a message and the probability of OK / NG can be determined in advance based on data accumulated by analyzing previous cases.

  FIG. 4A shows a description example of filtering a log in the log analysis unit. When the message shown in FIG. 4B is extracted by keyword detection, the NG probability of the node constituting the Bayesian network at the location (extraction location) in the right column in the table of FIG. Or set it to a value according to it. For example, as shown with reference numerals 41, 42, and 43 in FIG. 4 (a), description of the possibility of abnormality is performed for the three points of the NVM PWBA main body, the NVM PWBA connection unit, and the ESS PWBA main body. Is called. These pieces of information are failure information that is not reflected in the existing failure diagnosis information. Here, the existing failure diagnosis information refers to failure diagnosis information obtained by accumulating data of past failure cases. In this embodiment, PWBA is diagnosed by dividing it into a main body itself and a connection part.

  The description with the reference number 41 conveys a functional abnormality of the NVM as a message, and the description with the reference number 42 indicates the abnormality of the connection part of the ESS PWBA connected to the NVM, with a reference number 43. Has reported an abnormality that ESS PWBA cannot recognize NVM as a message. In this figure, it is written as if there is one abnormal part in one description. However, in actuality, a description over a plurality of parts is performed, or a plurality of information is included in one description. May be embedded.

  FIG. 5 shows an example of a Bayesian network configured based on the above-described concept for a PWBA mounted in the image forming apparatus 13. In the figure, reference numerals 51 and 52 indicate analyzed data relating to each node. That is, for example, consider the case where log data as shown in FIG. 4A is included in the analyzed data with reference numeral 51. As a result of analyzing the description content of Log1 that is the acquired log, when the content that is determined to be abnormal in the NVM PWBA main body indicated by reference number 41 or reference number 42 is found in FIG. The probability that an abnormality is recognized at the location, that is, the probability of becoming NG is set by the method described in the above. Similarly, as shown with reference numeral 43, when a content that is determined to be abnormal is found in the description of the ESS PWBA main body, it is created so as to increase the NG probability of the node.

  The allocation of probabilities as described above is performed in each node, and will be described below with reference to FIG. 6 taking NVM PWBA as an example. FIGS. 6A to 6D show the description contents of the specific probabilities of the respective nodes. Fault diagnosis is performed by sequentially following the probability allocation process from FIG. 6 (a) to FIG. 6 (d). In FIG. 6, (a) and (b) are probability distributions obtained from input information based on failure cases as in the conventional case. On the other hand, in FIG. 6, (c) and (d) are probability distribution tables that take into account the log description content, which is characteristic processing in the present invention.

  First, FIG. 6A will be described. FIG. 6A shows the failure probability of each component at the node of the failure code 001. When the value in the rightmost column of the table (NG value of the failure code 001) is not 0%, a part indicated as “NG” is an abnormality candidate and a failure cause candidate. As shown in FIG. 5, the fault code 001 is connected to ESS PWBA Total, MCU PWBA Total, and NVM PWBA Total. For this reason, in FIG. 6A, NVM PWBA Total indicating NG when the value in the rightmost column of the table is 5%, and MCU PWBA indicating NG when the value in the rightmost column of the table is 95%. Total is a candidate for abnormality.

  Next, FIG. 6B shows the failure probability at the node of the NVM PWBA Total that is an abnormal candidate from the probability table shown in FIG. Here, as shown in FIG. 5, since the NVM PWBA Total is connected to the NVM PWBA connection section and the NVM PWBA main body, it indicates the failure probability for these two nodes. When the value in the rightmost column of the table shown in FIG. 6B (the value of NG of NVM PWBA Total) is not 0%, a part denoted as “NG” is an abnormality candidate and a failure cause candidate.

  Next, FIG. 6C and FIG. 6D will be described. FIG. 6C shows that when the analyzed data obtained by analyzing the log includes a message indicating that there is a possibility that the NVM PWBA connection is abnormal, NG = 100% of the NVM PWBA connection is shown. Yes. Similarly, FIG. 6D shows that when the analyzed data obtained by analyzing the log includes a message indicating that there is a possibility that the NVM PWBA main body is abnormal, NG = 100% of the NVM PWBA main body is set. Yes. That is, in both cases, when a message indicating abnormality in the log description is found as a result of analysis, the probability of PWBA determined to be abnormal or NG of the connection unit is set to 100%. The assignment of these probabilities is performed based on past cases and is irrelevant to the assignment of probabilities shown in FIG. 6 (a) and FIG. 6 (b). That is, it is possible to perform failure diagnosis by assigning probabilities based on FIG. 6C and FIG. 6D, but in this embodiment, all of FIG. 6A to FIG. Failure diagnosis is taken into account, and the accuracy of failure diagnosis is improved.

  By creating the probability table for each node in this way, the final result as shown in FIG. 7 is obtained. In a normal case-based model in which logs are not acquired and a configuration in which parts with a possibility of failure are ranked and indicated, as described in the column of “failure display number 001 no log message” in FIG. Are displayed in the failure order of ESS PWBA, MCU connection cable, NVM PWBA, NVM connection unit, HDD main body,. In the present invention that has acquired the log, the description about the NVM indicated by reference numeral 41 is extracted by filtering as failure information that is not used in the log message, and “failure display number 001 has log message” is extracted. As shown in the column, the failure probability ranking is such that NVM PWBA is shown at the top with a high probability and greatly varies from the ranking using existing cases. In this way, by filtering and analyzing the log generated at startup and extracting messages that have not been used so far, the diagnostic accuracy in a system that performs failure diagnosis using existing cases is improved. It becomes possible.

  The message extraction will be described below. The log message content shown in FIG. 4 is extracted by comparison with a message output in a normal state registered in advance, or a specific message content is searched and extracted by a keyword, or the appearance of the searched keyword appears. When a plurality of keywords that reflect the position in the diagnosis result are extracted, the probability of NG at a predetermined location in FIG. 7 is set to 100% or a value corresponding thereto according to the order. A method is used such as comparing the value with that in a normal state or comparing a specific numerical value. Since logs created for the purpose of exchanging existing status information between control systems are used instead of logs created for so-called failure diagnosis, the extraction process according to the format of the message output there is Necessary. In addition, the storage location of case data, which is the base of the diagnostic model that reflects the analysis result of the log, will be mentioned. The case data may be stored in a memory mounted on each device, or may be retrieved by accessing a storage destination via a network. Alternatively, if the person who performs the check work individually responds to the site, prepare a PC to perform the work or a separate storage medium such as a memory, store it in it, and set it at the time of diagnosis for reference You may make it make it.

  As described above, the present invention is characterized by adding information based on a log created by the image forming apparatus to information for performing failure diagnosis in a failure diagnosis model using a Bayesian network or the like. Therefore, a conventional system can be used as the hardware configuration of the failure diagnosis system.

  The above-described embodiments are merely examples for carrying out the present invention, and the present invention is not limited to them. Various modifications of these embodiments are within the scope of the present invention. It is apparent from the above description that various other embodiments are possible within the scope. Further, in this specification, although described as a failure diagnosis system for an image forming apparatus, a log is created during startup of the apparatus, and data that can be used for failure diagnosis is extracted from this log. If present, the device can perform failure diagnosis using the failure diagnosis system of the present invention as a failure diagnosis target device.

It is explanatory drawing which showed the network to which the failure diagnosis system of this invention is applied, and the apparatus connected to it. 3 is a block diagram illustrating a connection state of an electronic circuit of the image forming apparatus. FIG. It is explanatory drawing which shows the flow which produces a probability table | surface from the log produced with the image forming apparatus. (A) is explanatory drawing which shows the example of a display of the extracted message, (b) is explanatory drawing which shows the example of a response | compatibility with a message and a failure location. It is explanatory drawing which shows the Bayesian network used for an Example. (A)-(d) is the probability table | surface which showed the failure probability allocated at each node, respectively. It is a result table | surface which shows an example of a failure diagnosis result.

Explanation of symbols

11 Desktop personal computer 12 Mobile personal computer 13 Image forming device 14 Network 21 ESS
22 IIT
23 IOT
24 FAX
25 MCU
26 HDD
212, 213, 214, 215, 216 Connecting cable

Claims (5)

A failure diagnosis system for an image forming apparatus,
A fault diagnosis unit that performs an inference process based on a fault diagnosis model that connects a cause of failure of the image forming apparatus and input information having a cause-and-effect relationship with the cause of the failure, and extracts anomaly candidates stochastically;
A log acquisition unit for acquiring a log created by describing status information of the image forming apparatus;
A log analysis unit that analyzes the log acquired by the log acquisition unit and creates analyzed data;
Have
The log analysis unit creates the analyzed data by extracting a predetermined keyword from the created log,
The failure diagnosis unit performs inference processing on the analyzed data in the log analysis unit as input information of the failure diagnosis model to which a failure probability value is assigned, and inputs the analyzed data based on the presence or absence of the keyword A failure diagnosis system characterized by changing the value of failure probability when inputting as information . A failure diagnosis system for an image forming apparatus,
A fault diagnosis unit that performs an inference process based on a fault diagnosis model that connects a cause of failure of the image forming apparatus and input information having a cause-and-effect relationship with the cause of the failure, and extracts anomaly candidates stochastically;
A log acquisition unit for acquiring a log created by describing status information of the image forming apparatus;
A log analysis unit that analyzes the log acquired by the log acquisition unit and creates analyzed data;
Have
The log analysis unit extracts a predetermined keyword from the created log and creates analyzed data including data related to a description position of the extracted keyword on the log,
The failure diagnosis unit performs inference processing on the analyzed data in the log analysis unit as input information of the failure diagnosis model to which a failure probability value is assigned, and based on the presence / absence of the keyword and the description position of the keyword A failure diagnosis system characterized by changing a value of a failure probability when inputting analyzed data as the input information . In the fault diagnosis system according to claim 1 or 2 ,
The failure diagnosis system, wherein the failure diagnosis unit, the log acquisition unit, and the log analysis unit are arranged outside the image forming apparatus and perform failure diagnosis via a network. An image forming apparatus comprising the failure diagnosis system according to claim 1 . A failure diagnosis method for an image forming apparatus, comprising:
A process of acquiring a log created by describing status information of the image forming apparatus;
The process of analyzing the log and extracting keywords on the log, and creating analyzed data including data related to the description position of the extracted keyword on the log;
The analyzed data is inferred by the failure diagnosis model as input information of a failure diagnosis model to which a failure probability value is assigned, and the analyzed data is input based on the presence / absence of the keyword and the description position of the keyword Changing the value of failure probability when entering as information ,
A failure diagnosis method comprising:

Images