JP2009003561A - Fault prediction diagnostic system and fault prediction diagnostic system using the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately predict the necessity/non-necessity of fault-coping to diagnostic object equipment. <P>SOLUTION: This fault prediction diagnostic device is provided with an information collection means 2 for collecting the internal information of diagnostic object equipment 10; a regression model creation means 3 for creating a regression model for calculating a fault prediction index corresponding to the fault prediction state of the diagnostic object equipment 10 based on the internal information of the diagnostic object equipment 10 collected by the information collection means 2; an index calculation means 4 for calculating a fault prediction index by using the regression model created by the regress model creation means 3; an index change digitizing means 5 for digitizing the secular change of the fault prediction index calculated by the index calculation means 4 to make it usable as a parameter; a fault-coping circumstance digitizing means 6 for digitizing the secular fault-coping circumstances to make it usable as a parameter; and a fault-coping determination means 8 for determining the presence/absence of the fault-coping of the diagnostic object equipment 10 based on each parameter of the index change digitizing means 5 and the fault-coping circumstance digitizing means 6. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a failure prediction diagnosis apparatus and a failure prediction diagnosis system using the same.

2. Description of the Related Art Conventionally, various failure prediction diagnosis systems that predict in advance a failure state of a diagnosis target device such as an image forming apparatus have been provided.
In Patent Document 1, a plurality of types of information related to the state of the image forming apparatus is acquired, and the Mahalanobis distance from the normal reference space is calculated using the acquired information, and the distance is calculated based on a predetermined threshold. Is larger, the technology for determining that the possibility of failure is high is disclosed.
In Patent Document 2, the jam code and the cumulative number of copies at that time are transmitted from the copying apparatus to the host computer, and the host computer obtains a cumulative jam curve based on the transmitted data, and performs a predetermined filtering process. A technique for inferring a jam risk level based on a filter output, a filter value, and a membership function of the risk level and determining whether or not to visit the user is disclosed.

Japanese Patent Laying-Open No. 2005-17874 (Embodiment of the Invention, FIG. 1) JP-A-10-161383 (Embodiment of the Invention, FIG. 14)

  The present invention is intended to provide a failure prediction diagnosis apparatus that accurately predicts whether or not a failure countermeasure is required for a diagnosis target device, and a failure prediction diagnosis system using the failure prediction diagnosis system.

  The invention according to claim 1 is a failure prediction diagnosis apparatus for diagnosing a failure prediction state of a diagnosis target device capable of storing internal information necessary for operation management, and collecting information on the internal information of the diagnosis target device A regression model creating means for creating a regression model capable of calculating a failure prediction index corresponding to the failure prediction state of the diagnosis target device based on the internal information of the diagnosis target device collected by the information collection unit, and the regression model An index calculation unit that calculates a failure prediction index using the regression model created by the creation unit, and an index change numerical value that quantifies the change over time of the failure prediction index calculated by the index calculation unit as a parameter. , Means for quantifying trouble handling status over time as a parameter, quantifying means for changing index, and number of troubleshooting status A failure prediction diagnosis apparatus characterized by comprising a failure address determining means for determining the necessity of the failure dealing diagnosis target device based on the parameters of the means.

According to a second aspect of the present invention, the failure prediction / diagnosis device according to the first aspect further comprises customer situation quantification means for digitizing the customer's proficiency level of the failure handling as a parameter so that the failure handling can be performed. A failure prediction unit, wherein the determination unit determines whether or not failure diagnosis of the diagnosis target device is necessary based on each parameter of the index change digitization unit, the failure handling status digitization unit, and the customer status It is a diagnostic device.
The invention according to claim 3 is the failure prediction diagnosis apparatus according to claim 1 or 2, wherein the failure prediction diagnosis apparatus is an image forming apparatus that forms an image on a recording material conveyed by the diagnosis target device. It is.
The invention according to claim 4 is the failure prediction diagnosis apparatus according to claim 1 or 2, characterized in that the failure prediction index is a failure risk level indicating a risk level to reach a failure of the diagnosis target device. It is.
The invention according to claim 5 is the failure prediction diagnostic apparatus according to claim 1 or 2, wherein the regression model creation means creates a regression model using logistic regression analysis. It is.

The invention according to claim 6 is the failure prediction diagnosis apparatus according to claim 1 or 2, wherein the regression model creation means uses internal information when the diagnosis target device fails as internal information used when creating the regression model; A failure prediction diagnosis apparatus including internal information after a failure is dealt with a failed diagnosis target device.
The invention according to claim 7 is the failure prediction diagnostic apparatus according to claim 1 or 2, wherein the regression model creating means updates the regression model.
The invention according to claim 8 is the failure prediction diagnosis apparatus according to claim 1 or 2, wherein the information collection means also collects environmental information as internal information.
The invention according to claim 9 is the failure prediction diagnosis apparatus according to claim 8, wherein the regression model creation means classifies the regression models created for each environmental category into a plurality of failure factors. This is a predictive diagnostic device.
The invention according to claim 10 is the failure prediction diagnostic apparatus according to claim 1 or 2, wherein the index change parameter quantified by the index change digitizing means is difference information from a predetermined threshold of the failure prediction index. The failure prediction diagnostic apparatus is characterized in that it is at least one of inclination change information of the failure prediction index and inclination change frequency information of the failure prediction index during a predetermined period.
The invention according to claim 11 is the failure prediction diagnosis apparatus according to claim 1 or 2, characterized in that the failure handling determining means determines the necessity of handling the failure using a Bayesian network probability inference engine. It is a failure prediction diagnosis device.
The invention according to claim 12 is the failure prediction diagnosis apparatus according to claim 1 or 2, further comprising a display unit capable of displaying the failure handling information determined by the failure handling determining unit. It is a diagnostic device.

According to a thirteenth aspect of the present invention, a plurality of diagnosis target devices capable of storing internal information necessary for operation management together with environmental information are connected to each diagnosis target device via a communication network that can communicate with each diagnosis target device. A failure prediction diagnosis system comprising a failure prediction diagnosis device for diagnosing a failure prediction state, wherein the failure prediction diagnosis device collects the internal information of the diagnosis target device, and the information collection unit A regression model creation means for creating a regression model capable of calculating a failure prediction index corresponding to the failure prediction state of the diagnosis target device based on the collected internal information of the diagnosis target device, and a regression created by the regression model creation means Index calculation means for calculating failure prediction index using a model, and change with time of failure prediction index calculated by this index calculation means can be used as a parameter Based on the parameters of the index change digitizing means, the failure handling situation digitizing means for digitizing the failure handling situation over time as a parameter, and the parameter change digitizing means and the failure handling situation digitizing means A failure prediction diagnosis system comprising failure handling determination means for determining whether or not failure diagnosis of a device to be diagnosed is necessary.
According to a fourteenth aspect of the present invention, in the failure prediction / diagnosis system according to the thirteenth aspect, the failure prediction / diagnosis device further quantifies the customer situation quantification so that the customer's proficiency level regarding failure handling can be used as a parameter. And the failure handling determination means determines whether or not failure diagnosis of the diagnosis target device is necessary based on each parameter of the index change digitizing means, the failure handling status digitizing means, and the customer situation. Is a failure prediction diagnosis system characterized by
The invention according to claim 15 is the failure prediction diagnosis system according to claim 13 or 14, wherein the information collecting means periodically collects internal information from each diagnosis target device.
The invention according to claim 16 is the failure prediction diagnosis system according to claim 13 or 14, wherein the information collection means collects information to be diagnosed for identifying the device to be diagnosed and internal information of the device to be diagnosed. It is a failure prediction diagnostic system characterized by collecting these together.

According to the first aspect of the present invention, it is possible to accurately predict the necessity of handling the failure for the diagnosis target device.
According to the second aspect of the invention, it is possible to consider the urgency of troubleshooting, taking into account the customer's proficiency level regarding troubleshooting, and to predict the necessity of handling the failure more accurately. it can.
According to the third aspect of the present invention, it is possible to accurately predict whether or not failure handling is required for the image forming apparatus as the diagnosis target device.
According to the invention which concerns on Claim 4, the failure risk as a failure prediction parameter | index of a diagnostic object apparatus can be estimated correctly.
According to the invention of claim 5, an accurate regression model can be easily created.
According to the invention of claim 6, it is possible to easily construct a regression model using past failure cases.
According to the invention which concerns on Claim 7, the influence of a season can be reflected effectively, and the necessity of failure handling can be predicted more correctly.
According to the invention which concerns on Claim 8, the failure state which tends to depend on environmental conditions can be estimated correctly.

According to the ninth aspect of the present invention, it is possible to accurately predict a failure state that tends to depend on environmental conditions for each failure factor.
According to the invention of claim 10, it is possible to accurately grasp the temporal change of the failure prediction index.
According to the invention of claim 11, it is possible to efficiently determine whether or not to deal with a failure.
According to the twelfth aspect of the present invention, it is possible to accurately grasp the result of necessity of troubleshooting.
According to the invention of claim 13, it is possible to accurately predict whether or not failure handling is necessary for a plurality of devices to be diagnosed.
According to the invention of claim 14, it is possible to consider the urgency of troubleshooting, taking into account the customer's proficiency level regarding troubleshooting, and accordingly, whether or not troubleshooting is necessary for a plurality of diagnosis target devices. Can be predicted more accurately.
According to the invention which concerns on Claim 15, the state of a some diagnostic object apparatus is always monitored, and a failure state can be estimated correctly.
According to the sixteenth aspect of the present invention, it is possible to specify a diagnosis target device and accurately predict a failure state while observing changes over time.

First, an outline of an embodiment model to which the present invention is applied will be described.
Outline of Embodiment Model FIG. 1 shows an outline of an embodiment of a failure prediction diagnosis system to which the present invention is applied.
In the figure, the failure reservation diagnosis system includes a plurality of diagnosis target devices 10 (for example, 10 (1), 10 (2), 10 (3)... 10 () that can store internal information necessary for operation management together with environmental information. n)), and a failure prediction diagnosis system 1 that is connected to the network 11 so as to be communicable with each diagnosis target device 10 and diagnoses a failure prediction state of each diagnosis target device 10. The predictive diagnostic apparatus 1 collects the internal information of the diagnostic target device 10 and the failure prediction state of the diagnostic target device 10 based on the internal information of the diagnostic target device 10 collected by the information collecting unit 2 A regression model creation means 3 for creating a regression model that can calculate a failure prediction index corresponding to the above, an index calculation means 4 for calculating a failure prediction index using the regression model created by the regression model creation means 3, The index change digitizing means 5 that digitizes the time-dependent change of the failure prediction index calculated by the index calculating means 4 as a parameter, and the fault countermeasure that digitizes the time-lapse failure countermeasure status as a parameter A situation digitizing means 6 and a failure handling determining means 8 for deciding whether or not to deal with a failure of the diagnosis target device based on the parameters of the index change digitizing means 5 and the failure handling status digitizing means 6 It is.

In such technical means, the diagnosis target is a failure prediction state of the diagnosis target device 10, and it is possible to predict and prevent a failure of the diagnosis target device 10 in advance.
Here, the 'internal information' may be information relating to one type of failure factor, but from the viewpoint of predicting failure prediction over a wide range, it is preferable to collect information relating to a plurality of types of failure factors.
In addition, when the diagnosis target device 1 is an image forming apparatus with high environmental dependency, it is preferable to capture environmental information as internal information. In this case, it is preferable to use temperature and humidity information as the “environment information”, but it is possible to use only the temperature or humidity information.
Further, as the information collecting means 2, when used in a failure prediction diagnosis system, a communication line with a communication network 11 connection is usually used, but the information collecting means 2 is not limited to this and is collected via a recording medium or the like. You may make it do.

Furthermore, as the regression model creation means 3, a regression model capable of calculating a failure prediction index may be created. Usually, internal information on past failure cases is used, or there is no past case of the same model. In such a case, it is sufficient to use internal information in a failure example of a related model. In addition, one regression model is not necessarily created, and a plurality of regression models may be created. In this case, one of the regression models may be selected.
Further, the failure handling determination means 8 widely includes a unit for determining whether or not failure handling is necessary according to a predetermined rule (inference engine) using the index change and the failure handling status as parameters.
Here, there may be one or a plurality of troubleshooting methods, and a representative method is to send a maintenance inspection specialist (for example, a service engineer), but this is not a limitation. It also includes methods such as consulting home with a TV conference system and telephone.

Further, in the present embodiment model, a typical aspect of the diagnosis target device 10 includes an image forming apparatus that has a high necessity for troubleshooting. The image forming apparatus here may be an image forming method or the presence or absence of a post-processing apparatus as long as it forms an image on a sheet to be conveyed.
Furthermore, as the “failure prediction index”, for example, a failure risk indicating the degree of risk leading to the failure of the diagnosis target device 10, a failure urgency indicating the urgency with respect to the failure of the diagnosis target device 10, and a symptom of the failure of the diagnosis target device 10 However, from the viewpoint of effectively preventing and predicting failure, it is preferable to use failure risk as a failure prediction index.

Further, from the viewpoint of maintaining good information collection efficiency in the information collecting means 2, a method of regularly collecting internal information together with environmental information from each diagnosis target device 10 is preferable.
Furthermore, from the viewpoint of specifying the diagnosis target device 10 and accurately predicting the failure state while observing the change over time, the information collecting unit 2 determines the diagnosis target information and the diagnosis target device 10 for specifying the diagnosis target device 10. It is preferable to collect the internal information and the information collection date and time information for collecting environmental information.

In addition, as the regression model creation means 3, a regression model may be set using a known regression analysis. As a typical method, for example, a regression model is created using logistic regression analysis.
Furthermore, from the viewpoint of easily constructing a regression model in the regression model creation means 3, internal information when the diagnostic target device 10 fails is used as internal information of the diagnostic target device 10 used when creating the regression model. It is preferable to use the information including the internal information after repairing the failed diagnostic target device 10.
Furthermore, from the viewpoint of more accurately reflecting the influence of the season, the regression model creating means 3 is preferably one that updates the regression model. In this aspect, the renewal time may be regular or may be appropriately selected such as a customer designated time.
Further, the regression model creation means 3 only needs to have at least one regression model. However, when an image forming apparatus having a high environment dependency or the like is used as the diagnosis target device 10, the regression model creation means 3 is created for each environment category. It is preferable to have a model, and from the viewpoint of performing diagnosis prediction more finely, the regression model for each environment category may be classified into a plurality of failure factors.

The index change parameter quantified by the index change digitizing means 5 may be selected as appropriate. However, the difference information with respect to a predetermined threshold of the failure prediction index, the inclination change information of the failure prediction index, and the predetermined It is preferable to use at least one, preferably all of the slope change count information of the failure prediction index during the period.
In addition, as a preferable aspect of the model of the present embodiment, the failure prediction diagnosis apparatus 1 further includes customer situation quantification means 7 that quantifies the customer proficiency status regarding failure handling as a parameter, The failure coping determination unit 8 determines whether or not the diagnosis target device 10 should be coping with the failure based on the parameters of the index change digitizing unit 5, the failure coping status digitizing unit 6 and the customer situation 7. It is done.
Furthermore, from the viewpoint of accurately grasping the necessity / unnecessary result of failure handling, a display means 9 capable of displaying the necessity / unnecessary result of the failure target determined by the failure handling determining means 8 may be provided. preferable. Of course, each parameter of the index change digitizing means 5, the failure handling situation digitizing means 6 and the customer situation 7 may be displayed on the display means 9.

Hereinafter, the present invention will be described in more detail based on embodiments shown in the accompanying drawings.
Embodiment 1
FIG. 2 shows the overall configuration of the failure prediction diagnosis system according to the first embodiment.
In the figure, the failure prediction diagnosis system can communicate with a plurality of image forming apparatuses 20 (specifically, 20 (1), 20 (2), 20 (3)... 20 (n)), respectively. A maintenance center 22 connected via a communication network 21 is provided.
In the present embodiment, the maintenance center 22 has a failure prediction diagnosis device 40 for diagnosing the failure prediction state of each image forming apparatus 20, and this failure prediction diagnosis apparatus 40 is managed by each image forming apparatus 20. A communication unit 41 for receiving information, a storage unit 42 for storing management information received by the communication unit 41, a regression equation as a regression model necessary for performing failure prediction diagnosis, and various input operations An input processing unit 43, an analysis processing / control unit 44 for controlling analysis processing for a failure prediction diagnosis request from the input unit 43, and a display unit 45 for displaying processing results by the analysis processing / control unit 44. I have.

In the present embodiment, as shown in FIG. 3, the image forming apparatus 20 forms an image on, for example, a sheet as a recording material in an image forming unit 31 such as an electrophotographic system. A storage unit 32 that stores processing programs and the like, a communication unit 33 that enables communication with the failure prediction / diagnosis device 40, and a control unit 34 that controls the image forming unit 31, the storage unit 32, and the communication unit 33 are provided. .
Further, the control unit 34 includes a temperature detector 36 for detecting the temperature inside or around the image forming apparatus 20, a humidity detector 37 for detecting the humidity inside or around the image forming apparatus 20, and a paper conveyance path. Are connected to a position detector 38 used for paper conveyance control and jam processing, and a paper counter 39 for counting the number of paper sheets.
The control unit 34 performs image forming processing by the image forming unit 31 in accordance with an image forming program stored in the storage unit 32 in advance, or management information necessary for operation management of the image forming apparatus 20 with respect to the storage unit 32. Is stored, or management information is transmitted to the failure prediction diagnosis apparatus 40.

As the management information described above, in addition to the internal information of the image forming apparatus 20, environmental information including temperature information from the temperature detector 36 and humidity information from the humidity detector 37, and further, a machine number of the image forming apparatus, And these data collection date information is used.
Here, supplementing the internal information of the image forming apparatus 20, there is a warning (warning) or a fail signal that the image forming apparatus 20 automatically sends depending on the operation state, each set value or observation value used for control, and the like. It is stored in a non-volatile memory as a storage unit 32 in the apparatus. For example, there are four types of fail signals: system failure, local failure, paper jam failure, and document jam failure. There are about 20-30 failures for each type. These are detected by various detectors (for example, the position detector 38) and a program provided in the image forming apparatus 20, and the content and number of failures and the history of failures that occurred in the past are also counted as the paper count when the failure occurs. The number is stored in the image forming apparatus 20 together with the number, and the latest fixed number of failures (for example, 20 for each type) is held.

FIG. 4 shows an example of internal information of the image forming apparatus 20.
In the figure, failure information indicates failure information that occurs in each part of the image forming apparatus 20.
IOT_LogicFail: Failure in the image output unit ESS_FanFail: Fan failure in the transmission subsystem (image signal generator) SoftFail: Failure in the software SensorC_Fail: Sensor failure USB_Open_Fail: Connection failure in the USB cable Communication Fail (for example, communication fail in the communication unit) Communication failure with image reader)
The jam information is jam detection information from the position detector 38 provided in each part of the sheet conveyance path.
Fuser_Jam: Jam at the fixing portion of the image forming unit 31 Regi_Jam: Jam on the alignment roll (registration roll) of the image forming unit 31 FeedOut_Jam: Jam at the paper supply unit Exit_Jam: Jam near the discharge roll TakeAway_Jam: Predetermined transport roll Further, the information on the current counter value−the counter value at the time of occurrence of abnormality is information indicating the number of sheets that have been normally imaged.

Further, in the present embodiment, the control unit 44 of the failure prediction / diagnosis apparatus 40 executes each process as shown in FIG.
(1) Management information request processing (see steps 51 to 53)
This periodically transmits a management information request signal to the image forming apparatus 20 to be diagnosed, and accordingly, receives management information from the image forming apparatus 20 side, and stores the received management information in the storage unit 42. To be stored.
(2) Regression formula creation (see step 54)
For example, as shown in FIG. 6, the internal information of the image forming apparatus 20 that has been requested for repair due to failure (collected as failure internal information) is collected together with temperature and humidity data. Thereafter, internal information of the image forming apparatus 20 that has been repaired and is operating normally (collected as normal internal information) is collected together with temperature and humidity data.
At this time, it is preferable to extract internal information and temperature / humidity data from a plurality of image forming apparatuses 20.
The temperature / humidity data is distributed, for example, in the range of 20 ° C. to 65 ° C. for temperature data and 124 to 135 (arbitrary relative value) for humidity data.
In the present embodiment, for example, as shown in FIG. 7, the temperature data area is divided into three and the humidity data area is divided into three and divided into nine environmental categories A to I.
In FIG. 7, one point indicates a plurality of image forming apparatuses having the same temperature and humidity. In this example, the central environmental section E includes approximately 50% of the image forming apparatuses.
In FIG. 7, using the failure internal information and normal internal information of the image forming apparatus falling into nine environment categories A to I, regression equations as regression models of nine failure risks for each of the environment categories A to I are obtained. create.
Here, for example, as shown in FIG. 8, a regression equation using logistic regression analysis is employed.
The created regression formula of failure risk is assigned an identification number and stored in the storage unit 42.

As described above, in this embodiment, a method of creating one regression model for each of the environmental classifications A to I may be adopted. However, from the viewpoint of performing a more detailed failure prediction diagnosis, the environmental classifications A to I are used. It is preferable to further categorize each failure factor in relation to each other and create a regression equation of failure risk for each category.
In other words, as described above, the internal information of the image forming apparatus in the past failure cases is used to create the regression formula of the failure risk, but the image forming apparatus that has failed in the past knows the failure location and the failure factor. So, for example, failure factors are divided into categories such as paper feeding troubles (jam troubles), image quality troubles, machine troubles, etc., and regression equations are created using the internal information of the image forming apparatus that caused each trouble. You can do it.

For example, in the case of a paper feed trouble, a regression equation of the failure risk of the paper feed trouble is created. The internal information related to the paper feeding trouble includes a paper jam failure, a document jam failure, a feed number interval at the time of failure, a local failure (failure related to a position detector for detecting the paper position), a consumable limit rate, and the like. Here, the limit rate of consumables means a value obtained by dividing the current use count by the limit use count of a roll, solenoid, motor, or the like used for paper feeding.
Therefore, the regression equation of the failure risk P (corresponding to D in FIG. 8) of the paper feed trouble is
Failure risk P = 1 / (1 + exp (−X))
However, X = A1 × (total number of paper jam failures) + B1 × (total number of document jam failures) + C1 × (average number of feeds between failures) + D1 × (total number of all paper feeds) + E1 × (consumption Product margin rate)
The coefficients A1 to E1 may be determined using internal information of past failure cases.
At this time, in the past failure cases of the image forming device, the internal information before the service engineer performs the maintenance check of the paper feed trouble and the internal information after the maintenance check (some internal information is Therefore, the coefficient of A1 to E1 (corresponding to K1 to K3 in FIG. 8) is obtained with P = 1 before maintenance inspection and P = 0 after maintenance inspection. This procedure is usually called logistic regression analysis.

Further, the regression equation of the failure risk degree G (corresponding to D in FIG. 8) of the image quality trouble is created using internal information categorized into image quality.
In other words, failure risk G = 1 / (1 + exp (−Y)) for image quality trouble
However, Y = A2 × (total number of system failures) + B2 × (total number of local failures) + C2 × (sensor measurement value related to image quality) + D2 × (average number of feeds between failures) + E2 × (consumables Marginal rate)
The consumable limit rate here is a consumable item related to image quality, and is a value obtained by dividing the current use count by the limit use count of drums, developers, and the like.
Then, the coefficients A2 to E2 (corresponding to K1 to K3 in FIG. 8) may be determined using internal information of past failure cases as in the case of the paper feed trouble.
Specifically, the values of A2 to E2 may be determined by using the logistic regression analysis method from the internal information before and after the day when the service engineer visited in the past due to the image quality trouble.
In addition, when there is little data regarding past internal information, calculation may not be possible as described above. Furthermore, since the characteristics of the image forming apparatus differ depending on the model, when there is a new new image forming apparatus introduced, there is no data regarding internal information of past failure cases, so data regarding internal information of other models is used. May be.

(3) Failure prediction diagnosis processing (step 55 to step 61 in FIG. 5)
This is a process when, for example, a service engineer requests failure prediction diagnosis for an image forming apparatus to be diagnosed.
At this time, the service engineer may input the machine number of the image forming apparatus to be diagnosed and the data collection date from the input unit 43, and the analysis processing / control unit 44 stores the corresponding internal data in the storage unit 42. Information, environmental information (temperature / humidity data) is extracted, and a regression formula for failure risk is selected based on the environmental information (temperature / humidity data). Internal information extracted along with the environmental information in the selected regression formula Is input, the failure risk based on the environmental information is calculated, and the value is displayed on the screen of the display unit 45 so that it can be referred to.
At this time, a plurality of regression equations of failure risk stored in the storage unit 42 are created for each of the environmental categories A to I, and are stored in the storage unit 42 together with the identification numbers. For this reason, if environmental information is known, it can respond to an identification number and a regression equation can be selected.

Further, the calculated failure risk is stored in the storage unit 42 together with the machine number, data collection date, and identification number of the selected regression equation, and the failure risk is displayed on the display unit 45 together with the machine number and data collection date. .
However, when the failure risk is already calculated from the machine number of the image forming apparatus to be diagnosed and the data collection date and stored in the storage unit 42, the failure risk is read from the storage unit 42 and displayed on the display unit 45. It is possible to display the failure risk along with the machine number and date of data collection.
Here, when the past history of failure risk is displayed in a graph, the machine number and the date range to be displayed may be designated from the input unit 43, and the above calculation is automatically repeated on the display unit 45. A graph corresponding to the past history can be displayed (for example, see FIG. 14).
In particular, as a regression formula of failure risk, in the case where the category is classified for each failure factor, if the failure risk of the paper feed trouble is greater than the failure risk of the image quality trouble, the paper feed trouble It is possible to determine that the risk of failure is high, and the value of the risk of failure can be calculated. This value can be used as a criterion for determining whether the service engineer needs maintenance or inspection when internal information is given (see, for example, FIG. 14).

At this time, the failure risk is displayed as a probability value between 0 and 1, and the failure prediction state of the image forming apparatus is indicated by the value.
If it is less than a predetermined threshold (for example, 0.5), it can be determined that the service engineer does not need to visit the customer. Further, for example, in the case of 0.5 or more and 0.6 or less, it is possible to watch the image forming apparatus as the diagnosis target because attention is necessary without visiting a customer. In addition, when a predetermined threshold value (for example, 0.6) is exceeded, a method in which a service engineer visits a customer and performs preventive maintenance even if the customer does not contact is generally used. It is a technique.

However, in determining whether or not preventive maintenance is necessary, not only the failure risk level of the image forming apparatus but also the urgency level indicating the level of the danger is required. The method of using only the failure risk threshold for the preventive maintenance determination is not preferable in that the urgency is not appropriately reflected. Appropriately reflecting the degree of urgency is a very important issue when deciding the order of priority for maintaining a large number of image forming apparatuses on the market and for efficiently utilizing a limited service engineer.
For example, when the risk of failure rises slowly over one month and exceeds the threshold, and when the risk of failure exceeds the threshold from almost zero in one day, the latter is highly urgent from past cases. It is clear even if it sees.
In addition, the urgency of dispatching a service engineer varies depending on the service engineer visit status and customer status. For example, it is clear that an image forming apparatus that a service engineer visits once every two months and an image forming apparatus that visits once every two weeks cannot be said to have the same degree of urgency even if the risk of failure is the same. .
Here, the customer status is a customer's proficiency level with respect to troubles in the image forming apparatus. Although it differs depending on the purpose of use of the image forming apparatus, even if the experienced customer who can easily clear the jam and the customer who is not the same, even if the same failure risk is displayed, it is not the same urgency. It is clear that immature customers are more urgent to dispatch service engineers.
Therefore, in the present embodiment, the change in failure risk over time, the visit status of the service engineer, and the customer status are parameterized, that is, converted into numerical values that can be used as parameters, for example, using the Bayesian network shown in FIG. A method of determining the necessity of dispatching a service engineer using a probabilistic inference engine is employed (see step 61 in FIG. 5).

-Service engineer dispatch necessity decision processing-
In this embodiment, the service engineer dispatch necessity determination process is performed by a Bayesian network probability inference engine based on failure risk level, service engineer visit status, and customer status parameters, as shown in FIG. 9, for example. , The probability of the necessity of dispatching the service engineer is obtained with reference to the conditional probability table incorporated in the table, and the necessity of dispatching the service engineer is determined based on the probability value.
Here, the situation of the failure risk is parameterized according to the rule shown in FIG. 10A, for example, the difference from the failure change threshold with respect to the threshold, the change in inclination, and the number of changes in the inclination during a certain period. Is.
The service engineer visit status is parameterized according to the rule of FIG. 10B, for example, the frequency of daily maintenance.
Further, the customer situation is parameterized according to the rules shown in FIG.

The nodes of the Bayesian network shown in FIG. 9 are seven: service engineer dispatch necessity, failure risk status, service engineer visit status, customer status, difference from threshold, slope change, and number of changes.
Here, the difference between the threshold, the change in slope, and the number of changes are the cause nodes of the failure risk status, and the failure risk status, service engineer visit status, and customer status nodes are service engineer dispatch necessity nodes It becomes the configuration which becomes the cause node.
The nodes of the difference from the threshold, the slope change, and the number of changes are data extracted from the failure risk parameter. Furthermore, past experience values are input as data to the service engineer visit status and customer status nodes. These data are stored in the storage unit 42 of the failure prediction / diagnosis device 40 of the maintenance center 22. These data are stored so that they can be searched from the machine number.
In the Bayesian network, as shown in FIG. 11, processing for determining whether or not a service engineer is dispatched is sequentially performed.
At this time, the nodes that calculate the probability in the Bayesian network are the failure risk level and the necessity of dispatching the service engineer. However, the value of the failure risk level is an intermediate node, so the result is unnecessary. The final required result is the probability of needing a service engineer. When the probability of necessity of dispatching a service engineer is calculated, if this value is larger than a preset threshold value (for example, 50%), a service engineer is dispatched. Judge that dispatch is unnecessary.

The storage unit 42 in the failure prediction diagnosis apparatus 40 stores data in which the code number assigned to the service engineer and the machine number are associated with each other, and when the service engineer inputs his / her code from the input unit 43. A list of machine numbers in charge can be displayed on the display unit 45 together with the risk of failure, and all the image forming apparatuses that need to be prevented from being inspected can be grasped (see, for example, FIG. 14).
Further, the storage unit 42 in the failure prediction diagnosis apparatus 40 stores data in which the area code and the machine number are associated with each other, and the code that can specify the area where the image forming apparatus is installed from the input unit 43. In other words, when the area code is input, a list of machine numbers in the area can be displayed on the display unit 45 together with the failure risk level, and it is possible to grasp all of the image forming apparatuses that require prevention of maintenance and inspection.

(4) Failure Risk Threshold Change Processing As shown in step 62 of FIG. 5, the failure risk threshold value has different characteristics depending on the model of the image forming apparatus, so that the threshold value can be changed after actual operation. It is preferable to do.

(5) Update regression equation (see step 54)
Since it is empirically known that the number of failures occurring in an image forming apparatus also changes depending on use conditions and environmental conditions, the regression equation for calculating the risk of failure needs to be updated periodically. . For example, jam failure is known to increase in the winter due to the relationship of humidity, so re-calculating the regression equation regularly (for example, every month) or irregularly when considering the influence of the season The method is more preferred. At this time, it is possible to change the failure risk threshold to cope with a change in the failure, but it is better to update the regression equation.

In the first embodiment, a regression model divided into a plurality according to environmental information is used. However, the present invention is not limited to this, and a single regression model may be used.
In the first embodiment, the customer engineer is taken into consideration in the process of determining whether or not the service engineer is dispatched. However, the present invention is not limited to this, and the customer situation may not be used.
Furthermore, a modified form of the failure prediction diagnosis system according to Embodiment 1 is shown in FIG.
In the figure, the failure prediction diagnosis system includes a communication network from not only the failure prediction diagnosis device 40 of the maintenance center 22 but also a portable information terminal 70 (for example, a notebook personal computer 70a, a PDA 70b, a mobile phone, etc.) owned by a service engineer. It is possible to collect information on the degree of failure risk via 21.
In this case, the service engineer may input his / her code to access the failure prediction diagnosis apparatus 40 and perform a series of failure prediction diagnosis processes.

FIG. 13 is a bar graph showing the number of times a service engineer visits a customer due to a paper feed trouble such as a jam in the image forming apparatus.
In the figure, there is a gradual increase over several years from the beginning of the introduction of the image forming apparatus from May XX to May ZZ.
This is considered a trouble due to wear.
Moreover, according to FIG. 13, periodicity is recognized according to the season. For example, the number of visits decreases from May to August, and increases from 11 to February. There is a difference of about twice between June and January.
This is considered to be due to the influence of ambient environmental information (temperature and humidity), which also affects the temperature and humidity in the image forming apparatus, and the number of occurrences of paper feeding troubles changes.

In such a state, a failure prediction diagnosis system as shown in the first embodiment, in particular, further classified into categories according to failure factors for each of the environmental classifications A to I, for example, failure risk of paper feed trouble / image quality trouble 14 was selected, and each failure risk was plotted over a predetermined number of days. As a result, the results shown in FIG. 14 were obtained.
In the figure, for example, when the failure risk exceeds 60% and the urgency is high, it is decided to dispatch a service engineer. Maintenance inspection work is performed by engineers.
In this situation, on the next day, since the maintenance inspection work by the service engineer is completed, the failure risk is 0.
As described above, according to the present embodiment, before a failure occurs, a service engineer can visit a customer to perform a maintenance inspection. In other words, the service engineer can make a visit plan and visit the customer efficiently and perform maintenance inspection, so that the service cost is greatly reduced, the downtime of the image forming apparatus is greatly reduced, and further, the customer It is understood that it leads to improvement of satisfaction.

Here, looking at the temporal change pattern of failure risk, for example, in the situation shown in FIG. 15, the difference from the threshold value is less than the threshold value, the slope change = 10% / day or more, less than 40% / day, during a certain period. It is understood that the dispatch of the service engineer is not necessary because the number of changes in the inclination of the system is 1.
In FIG. 16, the difference from the threshold is greater than or equal to the threshold, the slope change is 10% / day or more and less than 40% / day, the number of slope changes during a certain period = 2, and the risk of failure suddenly increases. Therefore, it is understood that the urgency of dispatching service engineers is high.
Further, in FIG. 17, since the difference from the threshold is less than the threshold, the slope change = 10% / day, the number of slope changes during a certain period = 1 or less, and the failure risk increases moderately, the service engineer is dispatched. It is understood that the urgency of is low.

Furthermore, as shown in FIGS. 18 (a) and 18 (b), it was also confirmed that the responsiveness of the change over time in the failure risk in this example is better than the filter output of the comparative example.
In this comparative example, the FIR filter (Finite Impulse Response Filter) of Patent Document 2 is used as a high-pass filter, and the filter output is calculated under the following filter conditions. It is understood that the slope of the part is gentle and the peak is delayed by about ½ of the filter length.
However, the filter condition of the comparative example, q (filter length) = 9, enter the _{x t,} outputs _{y t,}
a _j (filter coefficient) = (a ₀ , a ₁ , a ₂ ... a _q )
= (-0.3, -0.225, -0.15, -0.075,0,0.075,0.15,0.225,0.3), it is expressed by the following formula 1.

It is explanatory drawing which shows the outline | summary of the failure prediction diagnostic system of embodiment model to which this invention is applied. It is explanatory drawing which shows the failure prediction diagnostic system which concerns on Embodiment 1. FIG. 1 is an explanatory diagram showing an overview of an image forming apparatus used in Embodiment 1. FIG. 2 is an explanatory diagram illustrating an example of internal information of an image forming apparatus used in Embodiment 1. FIG. It is explanatory drawing which shows the operation | movement process of the failure prediction diagnostic system which concerns on Embodiment 1. FIG. It is a flowchart which shows an example of the regression type preparation process of FIG. 6 is an explanatory diagram illustrating an example of an environment classification of a plurality of regression equations used in the first embodiment. FIG. It is explanatory drawing which shows an example of the logistic regression analysis used when producing the regression type used in Embodiment 1. FIG. It is explanatory drawing which shows an example of the inference engine for embodying the service engineer dispatch necessity determination process of FIG. (A)-(c) is explanatory drawing which shows the example of a rule about the parameterization of the situation of failure risk of FIG. 9, the visit situation of a service engineer, and a customer situation. It is a flowchart which shows an example of the service engineer dispatch necessity determination process of FIG. It is explanatory drawing which shows the failure prediction diagnostic system which concerns on the modification of Embodiment 1. It is explanatory drawing which shows an example of the relationship between the visit date and the number of visits of the service engineer in an Example. It is explanatory drawing which shows the example of a change of the failure risk in an Example. It is explanatory drawing which shows the change pattern 1 of failure risk. It is explanatory drawing which shows the change pattern 2 of failure risk. It is explanatory drawing which shows the change pattern 3 of failure risk. (A) (b) is the graph which contrasted the change pattern of the failure risk in an Example, and the change pattern of the filter output in a comparative example.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Failure prediction diagnostic apparatus, 2 ... Information collection means, 3 ... Regression model preparation means, 4 ... Index calculation means, 5 ... Index change numerical conversion means, 6 ... Failure handling situation numerical conversion means, 7 ... Customer situation numerical conversion means , 8... Failure handling determination means, 9... Display means, 10 (10 (1) to 10 (n))... Diagnosis target device, 11.

Claims (16)

A failure prediction diagnosis device for diagnosing a failure prediction state of a diagnosis target device capable of storing internal information necessary for operation management,
Information collecting means for collecting the internal information of the device to be diagnosed;
A regression model creation means for creating a regression model capable of calculating a failure prediction index corresponding to the failure prediction state of the diagnosis target device based on the internal information of the diagnosis target device collected by the information collection means;
Index calculation means for calculating a failure prediction index using the regression model created by the regression model creation means;
Index change digitizing means for digitizing the change over time of the failure prediction index calculated by the index calculating means so that it can be used as a parameter;
Failure handling status quantification means for digitizing the failure handling status over time as a parameter,
A failure prediction diagnosis apparatus comprising failure handling determination means for determining whether or not failure handling of a diagnosis target device is necessary based on each parameter of the index change digitizing means and the failure handling status digitizing means. In the failure prediction and diagnosis apparatus according to claim 1,
Furthermore, a customer situation quantification means is provided for quantifying the customer's proficiency status regarding troubleshooting as a parameter,
The failure handling determination means determines whether or not failure diagnosis of the diagnosis target device is necessary based on each parameter of the index change digitizing means, the failure handling status digitizing means, and the customer situation. Failure prediction diagnosis device. In the failure prediction diagnostic apparatus according to claim 1 or 2,
A failure prediction diagnosis apparatus, wherein the diagnosis target device is an image forming apparatus that forms an image on a recording material to be conveyed. In the failure prediction diagnostic apparatus according to claim 1 or 2,
The failure prediction index is a failure risk level indicating a risk level that leads to a failure of the diagnosis target device. In the failure prediction diagnostic apparatus according to claim 1 or 2,
A failure prediction diagnosis apparatus characterized in that the regression model creation means creates a regression model using logistic regression analysis. In the failure prediction diagnostic apparatus according to claim 1 or 2,
The regression model creation means includes internal information when the diagnosis target device has failed and internal information after troubleshooting the failed diagnosis target device as internal information used when creating the regression model. A failure prediction and diagnosis apparatus characterized by the above. In the failure prediction diagnostic apparatus according to claim 1 or 2,
A failure prediction diagnosis apparatus characterized in that the regression model creation means updates the regression model. In the failure prediction diagnostic apparatus according to claim 1 or 2,
A failure prediction diagnosis apparatus characterized in that the information collecting means also collects environmental information as internal information. The failure prediction and diagnosis apparatus according to claim 8,
A failure prediction diagnosis apparatus characterized in that the regression model creation means is a regression model created for each environment category, which is classified into a plurality of failure factors. In the failure prediction diagnostic apparatus according to claim 1 or 2,
The index change parameter quantified by the index change digitizing means includes information on the difference between the failure prediction index and a predetermined threshold, slope change information on the failure prediction index, and information on the number of slope changes of the failure prediction index during a predetermined period. It is at least one of these, The failure prediction diagnostic apparatus characterized by the above-mentioned. In the failure prediction diagnostic apparatus according to claim 1 or 2,
A failure prediction diagnosis apparatus characterized in that the failure handling determination means determines whether or not a failure handling is necessary using a probability inference engine based on a Bayesian network. In the failure prediction diagnostic apparatus according to claim 1 or 2,
A failure prediction diagnosis apparatus comprising display means capable of displaying the failure handling information determined by the failure handling determination means. Failure prediction that diagnoses the failure prediction status of each diagnosis target device by connecting to multiple diagnosis target devices that can store internal information necessary for operation management together with environmental information and a communication network that can communicate with each diagnosis target device A failure prediction diagnosis system comprising a diagnosis device,
The failure prediction diagnostic apparatus is
Information collecting means for collecting the internal information of the device to be diagnosed;
A regression model creation means for creating a regression model capable of calculating a failure prediction index corresponding to the failure prediction state of the diagnosis target device based on the internal information of the diagnosis target device collected by the information collection means;
Index calculation means for calculating a failure prediction index using the regression model created by the regression model creation means;
Index change digitizing means for digitizing the change over time of the failure prediction index calculated by the index calculating means so that it can be used as a parameter;
Failure handling status quantification means for digitizing the failure handling status over time as a parameter,
A failure prediction diagnosis system comprising failure handling determination means for determining whether or not failure handling of a diagnosis target device is necessary based on each parameter of the index change digitizing means and the failure handling status digitizing means. The failure prediction diagnosis system according to claim 13,
The failure prediction diagnosis device further includes a customer situation quantification means for quantifying the customer's proficiency status regarding failure handling as a parameter,
The failure handling determining means determines whether or not failure handling of the diagnosis target device is necessary based on each parameter of the index change digitizing means, the failure handling status digitizing means, and the customer situation. Failure prediction diagnosis system. The failure prediction diagnosis system according to claim 13 or 14,
A failure prediction diagnosis system, wherein the information collection means periodically collects internal information from each diagnosis target device. The failure prediction diagnosis system according to claim 13 or 14,
The information collection means collects together diagnosis target information for identifying a diagnosis target device and information collection date and time information for collecting internal information of the diagnosis target device.

Images