JP2693302B2 - Image forming apparatus having self-repair system - Google Patents

Abstract

PURPOSE:To effectively perform the self-recovery to a failure by use of a case in which recovering work is described. CONSTITUTION:Cases CS1-CS5 are hierarchically classified according to failure symptoms, failures and classes A-C, and a case suitable to a caused failure is selected to perform recovering work. Application priorities are given to the classes A-C, and when the applicable case is present in plural classes, the application priorities are referred to at need. When the application of the case is successful, the application priority of the class to which the case belongs is raised to make this class easy to apply. When the failure can not be recovered by the case application, the work is inferred by qualitative inference to recover the failure. When it is successful, the inferred work is registered as the case of a new class.

Description

DETAILED DESCRIPTION OF THE INVENTION

Contents 1. Industrial application field 2. Conventional technology 3. 3. Problems to be Solved by the Invention Means for solving the problems 5. Action (5-1) Means for solving the problem (5-2) Contents and action of "detailing work script" 6. Example (6-1) Outline of system configuration (6-2) Outline of system operation (6-3) Specific configuration and state of target machine (6-4) Entity model and mathematical model (6-5) Failure diagnosis method (6-6) Execution of repair work (6-7) Failure repair processing by case retrieval (6-8) Reasoning for repair plan (6-9) Reasoning for side effects (6-10) Other 7. The invention's effect

[0002]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image forming apparatus having a self-repair system. More specifically, the present invention relates to an image forming apparatus capable of self-repairing (self-diagnosing an operating state and the like) by utilizing artificial intelligence and knowledge engineering, which have been actively researched in recent years.

[0003]

2. Description of the Related Art In the field of development of precision machinery and industrial machinery, artificial intelligence (ArtificialInt) has recently been developed in order to save labor in maintenance work and prolong automatic operation.
Elligence: Research on expert systems using so-called AI) technology is being actively conducted. Some expert systems perform self-diagnosis as to whether or not a failure has occurred in the device and self-repair the resulting failure.

[0004]

However, the conventional expert system (automatic adjustment system or failure diagnosis system) basically operates the corresponding actuator based on the output of a certain sensor. . That is, a kind of automatic adjustment and failure diagnosis have been performed by a combination of a predetermined sensor and actuator.
Therefore, basically, a certain sensor corresponds to a specific actuator, and the relationship between the two is fixed. Therefore, (1) The relationship between the parameters of the sensor and the parameters of the actuator must be specified numerically.

(2) For the reason of (1) above, the relationship between the sensor parameter and the actuator parameter strongly depends on the object, has poor versatility, and cannot be used for various objects. (3) The relationship between the parameters of each sensor or the parameter of each actuator is irrelevant to the control.
Therefore, only simple control based on the relationship between the corresponding sensor parameter and actuator parameter can be performed, the failures that can be dealt with are limited in advance, and unknown failures cannot be handled.

(4) For the reason of the above (3), it is impossible to predict a secondary effect on other actuator parameters which may be caused by operating an arbitrary actuator.
And so on. As described above, in the conventional automatic adjustment system and failure diagnosis system, the predicted failure A is performed based on the set A of the sensor A and the actuator A, and the predicted failure B is performed based on the set B of the sensor B and the actuator B. In other words, the predicted failure C is performed based on the set C of the sensor C and the actuator C, and the failure diagnosis is performed based on the independent set of the sensor and the actuator C, and the failure is repaired based on the diagnosis. Did not.

[0007] Therefore, the present applicant has filed a patent application for a new self-diagnosis and self-repair system for an image forming apparatus which overcomes the disadvantages of the prior system (e.g. Japanese Patent Application No. Rights 2-252191 (JP-A-4 -130459
No.)) ). The self-diagnosis and self-repair system for the image forming apparatus according to this prior application has two major features.

[0008] One is to qualitatively represent a target machine using parameters, and to perform fault diagnosis of the target machine using the qualitative data, that is, a qualitative model based system (qualitative model based).
System: Hereafter, the failure diagnosis is performed by "QMS". And the other is QMS
When the processing is performed, the result of the failure diagnosis is hierarchically classified by the “failure symptom” and the “failure”, and the result is stored as an example. In addition, since a common repair work can often be performed on a plurality of cases that belong to the same failure symptom and failure, the work required for repair is expressed in a rule form with a minimum unit. Then, the set of work units is registered as a work script and shared by the failure symptom and a plurality of cases belonging to the failure. The case-based restoration planning system (Case Based Planning System) is used by using the case base thus obtained.
em: Making "CBS"). As a result, from the cases hierarchically classified by the failure symptom and the failure, the case to which this failure symptom and the failure apply is selected, and the repair work is performed based on the selected case and the work script corresponding to the selected case. Be seen.

By the way, in the invention of the prior application, the cases in CBS are classified by "fault symptom" and "fault", but it seems that the cases can be further classified. By further classifying cases, when referring to a case, if reference to a case fails, avoid selecting a case that belongs to the same classification as the failed case, and improve the success rate for case selection. It should be possible to improve.

The inventor of the present application has completed the present invention by paying attention to the point of detailing the case classification. In short, the object of the present invention is to provide an image forming apparatus having a system that improves the success rate regarding case selection in CBS and can perform self-repair work more quickly.

[0011]

In order to realize the above-mentioned object, the image forming apparatus of the present invention stores the cases in which the work for repairing the failure of the image forming apparatus is described by classifying the cases. When a failure occurs in the storage unit, the priority storage unit that stores the application priority for each type of failure repair, and the image forming apparatus, when a case corresponding to the failure is selected and there is an item with the same degree of correspondence Selects a case in the order of the highest application priority and applies the method to the failure repair, and the application priority of the class to which the case belongs stored in the priority storage means according to the application result of the case. Updating means, and a work inference means for inferring a work for repairing the failure by using qualitative data predetermined with respect to the image forming apparatus when the failure repair by applying the case is not successful, and the work inference means Work inferred by Therefore, when a successful fault recovery,
A new class framework is created in the case storage means, the inferred work is registered as a case belonging to the new class, and the application priority of the new class is set to the highest, and the application priorities of other classes are adjusted accordingly. And means for updating.

[0012]

[Action]

(5-1) Action of Means for Solving the Problem In the image forming apparatus according to the present invention, cases in which the work for repairing a failure is described are classified and stored. Categorization is based on whether cases are qualitatively within reach. In addition, application priorities for each type of failure repair are stored.

When a failure occurs in the image forming apparatus, first, a case corresponding to the failure, for example, a case having a high degree of coincidence of state parameters is selected. If there are cases with the same degree of correspondence, the cases are selected in order of the highest application priority and applied to failure repair. If the repair work is successful as a result of applying the selected case to the failure repair, the application priority of the class to which the applied case belongs is updated. On the other hand, when the repair work fails, the application priority of the class to which the applied case belongs is updated to be low. If the repair work is not successful as a result of applying the first selected case with the highest application priority, the next case with the next highest application priority is selected and applied to the failure repair. It As a result, if the failure repair is successful, the application priority of the class to which the case belongs is updated. As a result, the application priority is highest for the case where the repair work succeeded by applying this time, and the application priority is set so that the case belonging to that class is selected first at the next failure repair. Will be updated.

When the repair work is not successful depending on the case application, the failure repair work is inferred by so-called qualitative inference using the qualitative data. If the inferred repair work succeeds in the failure repair, a new class of framework is created in the case storage means, and the inferred repair work is newly registered as a case belonging to the new class. While the repair failed in the case applied prior to qualitative reasoning,
According to the work inferred using the qualitative inference, the restoration has succeeded, so it can be said that the previous case and the case related to the inferred work have a qualitatively unreachable relationship. Therefore, the case related to the inferred work is registered as a case belonging to a new class. In addition, this new class is prioritized to have the highest application priority,
The application priorities of other types are updated accordingly.

(5-2) Content and Function of "Detailing of Work Script" The image forming apparatus according to the present invention basically has a CBS. Regarding the basic structure of CBS,
Although disclosed in the prior application of the applicant of the present application, it will be described in detail in the examples of the present specification as well. C
In the BS, as described above, a "work script" described in a rule format is defined as a method of expressing the operation of the minimum unit of the actual repair work. Then, an operation of “detailing the work script” is introduced for the purpose of expanding and interpreting the case. Refinement of the work script is to add a logical operation by hypothetical reasoning to the work script, find the difference between the case situation and the current situation, and eliminate the difference to determine the current situation. It is an operation to make qualitatively equal to the situation of the case.

A more specific description will be given. As shown in FIG. 1, for example, in a two-dimensional space represented by parameters A and B, the inside of a circle indicated by FS is a parameter space in which the device is in a normal state. And case 1
In the case of, the state position represented by parameter A and parameter B is CS1, and a failure has occurred. Therefore, by performing the repair work C1R described in the work script, the state position CS1 of Case 1 moves into the normal parameter space FS, and the repair work succeeds.

In case 2, the state position represented by the parameters A and B is CS2, and the repair work C2 described in the work script for this state is C2.
By applying R, this state position can be moved into the normal parameter space FS. The repair work C1R described in the work script is a repair work that is effective when the status of the case is at the position indicated by CS1. Similarly,
The repair work C2R described in the work script is a valid repair work when the status of the case is the position of CS2.

By the way, when the newly diagnosed current failure situation is the situation indicated by CS3, neither the repair work C1R nor C2R described in the work script can be applied as it is. Therefore, the operation C3R1 for finding the difference between the current situation CS3 and the situation CS1 of case 1 and moving CS3 to the situation of CS1 is added to the work script. This is the detailed work script. Alternatively, repair work C3R2 for moving the current situation CS3 to the situation CS2 of case 2 is added. This is the detailed work script. Then, if the current situation CS3 can be moved to the situation CS1 of the case 1, for example, the repair work C1R can be applied to the situation to move the device into the normal parameter space FS.

Now, the present situation CS3 is repaired C3R1
If it is possible to move to the situation CS1 of the case 1 by performing, and if it can be moved to the parameter space FS by performing the repair work C1R, it is possible to refer to the case by performing the refinement of the work script. It means that it succeeded in. Therefore, in this case, the situation in which the case is successfully referred to by the detailed work script, that is, the current situation CS3 and the original case, that is, the situation CS1 of the case 1 are within the same qualitatively reachable range. I can think. Cases within the same qualitatively reachable range are defined as belonging to the same class.

On the other hand, the current situation CS3 is changed to the situation C of case 2.
When it tried to move to S2, but it did not succeed, the current situation CS3 and the situation CS2 of case 2 are qualitatively unreachable. therefore,
Those in this relationship are defined as belonging to different classes. Further, in a certain situation, a case where reference fails and a case where reference succeeds are qualitatively unreachable and are defined as belonging to different classes. Also, I failed to refer all cases,
Cases newly generated by QMS are qualitatively unreachable from previous cases and are defined as belonging to a different class.

In the CBS according to the prior invention of the applicant of the present application, as shown in FIG. 2, the cases are hierarchically separated by the "fault symptom" and the "fault". As a result of the definition, in the present invention, as shown in FIG. 3, the cases are hierarchically classified in more detail by “fault symptom”, “fault” and “class”. Next, a specific example of case classification will be shown.

Description will be made with reference to FIG. As an example, C
It is assumed that there are five cases from S1 to CS5. Among them, CS1, CS3 and CS4 are repair cases generated by QMS, and CS2 is from CS1 to CS.
It is assumed that each of the cases 5 is obtained from the CS 4 through the detailing by the case inference. In this case, the five cases are [CS1, CS2], [CS3], [CS4, CS5].
It is classified into three categories.

Now, when referring to these cases in an arbitrary situation S, the order regarding case reference is given in the order in which the qualitative distance between the situation S and the situations of these cases seems to be short. The ranking result is CS2, CS1,
It is assumed that the order is CS4, CS3, CS5. At this time, first, the case CS2 is referred to, and if the reference fails, the case CS1 is not referred to. This is based on the fact that the reason why the case reference fails is that the given situation S is judged not to be within the qualitative reachable range of the case CS2, and the case CS1 belonging to the same class is also considered to be the same. ing. Therefore, if all the case references fail, the references will be made in the order of CS2, CS4, CS3.

In this way, by introducing the concept of "class" into the classification of cases and refining the classification of cases, it is possible to increase the success rate regarding the selection of cases to be referred to.
In addition, it is possible to shorten the time for selecting cases. By the way, as shown in FIG. 3, in the present invention, the case sets are classified into classes. Then, cases belonging to the same class are within a qualitatively reachable range. Therefore, for example, the case CS1 and the case CS4 are qualitatively unreachable. Therefore, there are qualitatively unreachable cases (eg CS1 and CS4).
It is unnatural for two to share the same set of work scripts. Therefore, in the present invention, different work script sets are assigned to case sets belonging to different classes. That is, as shown in FIG. 5, the work script set is shared for each qualitatively reachable case set.

From a different point of view, it can be considered that the root causes of failures are different between cases having a qualitatively unreachable relationship, and therefore work is performed for each case belonging to a common class. Multiplexing the work script sets to allocate the script sets can be regarded as the work of separating the work script sets according to the root cause of the failure. Further, in the present invention, in order to improve the efficiency of case retrieval, a process of advancing the priority order of case application based on the similarity of the classes is applied to the cases having the same failure symptom and failure but different classes. ing.

Each case is hierarchically classified by "failure symptom" and "failure" which are the results of QMS failure diagnosis. Then, in the process of restoration by case reasoning being repeated, it is further classified into "classes". Therefore, as a result, the work of classifying cases by "class" is further classified into "faults" in the failure diagnosis.

By the way, the "class" is a relationship between states that are qualitatively unreachable positions as described above, and can be considered as a difference in the root cause of a failure. However, considering an actual mechanical system, the root cause of a failure is often the same, but the manifested failure symptoms often differ. For example, taking a copying machine as an example, there is a case where “a fog occurs when the main charge rises, but the drum may leak and reset.” Or “a fog occurs when the main charge rises. In some cases, there is a case where "the image leaks into the drum and the image is disturbed." . Furthermore, the device has a high frequency of failure symptoms that are manifested by the same failure cause unless the root cause of the failure is eliminated.

Therefore, by applying a case of a certain class with respect to a certain failure symptom and succeeding in repairing, if a case of a similar class is applied with respect to a different failure symptom,
It is highly likely that appropriate repair work will be obtained.
Therefore, if a case of a certain failure symptom is applied and repaired, and if it succeeds, there is a case belonging to a different failure symptom and a case similar to the case where the repair is performed. , The process of raising the priority of application of the case is performed.

In other words, the classification by "class" is "beyond the framework" of the QMS-based classification of "fault symptom" and "fault", so that "class" is similar even in cases relating to different fault symptoms. There is a case. here,
The similarity of “class” means that cases belonging to the class have common characteristics for each class. As will be described in detail later in Examples, the feature extraction of the class is performed by extracting the common part from the “state before repair” of the cases belonging to the class. Cases belonging to different failure symptoms are compared with each other, and when the “status before repair” of all the cases included in the class have the same characteristics, it is determined that they are similar, and the priority order for case application Carry forward processing.

For example, as shown in FIG. 6, when the characteristic of the case set A is P and the characteristic of the case set B is Q, and when P is included in Q (PεQ), the case set A is It is determined that it is similar to the case set B. In this case, the case set B is not always similar to the case set A (Q is QεP included in P). By advancing the application order of cases by this kind of similarity, it is possible to improve the overall success rate of case application in case-based restoration.

[0031]

【Example】

(6-1) Outline of System Configuration FIG. 7 is a block diagram showing the system configuration of an embodiment of the present invention. This system includes a plurality of sensors 1a, 1b, 1c installed on an image forming apparatus which is a target machine.
And a plurality of actuators 6a, 6b, 6c for changing the functional state of the target machine.

Each of the plurality of sensors 1a, 1b, 1c is for detecting a change in an element of the target machine or a related state between the machine elements caused by the operation of the target machine. The information fetched from each of the plurality of sensors 1a, 1b, 1c is amplified by the amplifier circuit 2, converted from an analog signal to a digital signal by the A / D conversion circuit 3, and given to the system control circuit 10.

The system control circuit 10 includes a digital signal / symbol conversion unit 11, a failure diagnosis unit 12, a failure simulation unit 13, a target model storage unit 14, a repair planning unit 15, and a symbol / digital signal conversion unit 16. There is. Further, a case base storage unit 17 and a work script storage unit 18 are connected to the repair planning unit 15.

The digital signal / symbol converter 11
The digital signal supplied from the A / D conversion circuit 3 is converted into qualitative information. That is, a conversion function is provided for converting a digital signal into one of three symbols, for example, normal, high, and low. Converting the signals provided by the sensors 1a, 1b, 1c into such symbolized qualitative information facilitates an approach to fault diagnosis. The symbols are not limited to normal, high, and low as in this example, and may be on and off or other expressions such as A, B, C, and D.
When the conversion unit 11 converts a digital signal into a symbol, the characteristic data unique to the target machine stored in the target model storage unit 14 is referred to. Details of the feature data and signal conversion will be described later.

The failure diagnosis section 12 and the failure simulation section 13 compare the symbols converted by the digital signal / symbol conversion section 11 with the failure diagnosis knowledge stored in the target model storage section 14 to determine whether or not there is a failure. It is a component that discriminates and performs a fault diagnosis, and as a result, represents the fault state of the target machine by qualitative information, that is, a symbol, and outputs it.

The repair planning unit 15, the case-based storage unit 17, and the work script storage unit 18 are
It is a component for inferring a repair plan and deriving a repair work based on the “failure symptom” and the “failure” that are the results of the failure diagnosis. In inferring the repair plan and deriving the repair work, a case regarding past repair success stored in the case base storage unit 17 is searched, and a work script for executing the searched successful case (minimum repair work). A unit operation is expressed and a series of work units for performing the repair operation described in the rule format; details will be described later.) Is selected from the work script storage unit 18.
Further, the qualitative data (described in detail later) stored in the target model storage unit 14 is utilized.

The restoration work output from the restoration planning unit 15 is converted into a digital signal in the symbol / digital signal conversion unit 16 with reference to the storage information of the target model storage unit 14. The digital signal is D /
The signal is converted into an analog signal by the A conversion circuit 4 and given to the actuator control circuit 5. Actuator control circuit 5
Is a plurality of actuators 6a, 6a based on an analog signal provided, that is, an actuator control command.
b and 6c are selectively operated to execute the repair work.

(6-2) Outline of system operation FIG.
8 is a flowchart showing a process of system control circuit 10 in FIG. 7. Next, with reference to FIG. 8, an outline of processing of the system control circuit 10 of FIG. 7 will be described. The detection signal of the sensor 1a, 1b or 1c is amplified and converted into a digital signal, which is read into the system control circuit 10 for every predetermined read cycle (step S1).

The read digital signal is symbolized by the digital signal / symbol converter 11 (step S2). This symbolization is performed based on the characteristic data preset in the target model storage unit 14, that is, the reference value data peculiar to the target machine. For example, in the target model storage unit 14, the output ranges of the sensors 1a, 1b, 1c are set as follows as reference value data specific to the target machine.

[0040] That is, the sensor 1a: output ka ₁ below = low output ka ₁ ~ka ₂ = exceeded normal output ka ₂ = high sensor 1b: less than output kb ₁ = low output kb ₁ ~kb ₂ = normal output kb ₂ Excess = High Sensor 1c: Output less than kc ₁ = Low output kc _{1 to} kc ₂ = Normal output Exceeds kc ₂ = High. Digital signal / symbol converter 1
In No. 1, the read digital signal is converted into a symbol “low”, “normal” or “high”, respectively, based on the reference value data specific to the target machine set in the target model storage unit 14. .

Next, in the failure diagnosis unit 12, the converted symbols are evaluated, the presence / absence of a failure is determined, and the failure symptom is specified (step S3). The failure diagnosis knowledge stored in the target model storage unit 14 is utilized for determining the presence / absence of a failure and identifying the failure symptom by evaluating the symbol. The fault diagnosis knowledge is, for example, a setting condition that a specific parameter must be, for example, normal. If the specific parameter is not normal, it is determined that there is a failure, and the failure symptom is specified by what the specific parameter is. If there is no failure, the routine of steps S1, S2 and S3 is repeated.

If it is determined that there is a failure in step S3, the state of the target machine is inferred, that is, failure diagnosis and failure state simulation are performed (step S4). Specifically, in the failure diagnosis unit 12, based on qualitative data stored in the target model storage unit 14 that qualitatively represents the behavior or attribute of each element forming the device and the connection relationship between each element. , The parameter causing the failure is retrieved, and the failure simulation unit 13
In, the failure condition is simulated assuming that the retrieved parameter is the cause of the failure. Further, the fault diagnosis unit 12 compares the simulation result with the current parameter value, and judges the validity of the assumption that the retrieved parameter is the cause of the fault.
The above processing is performed for all of the plurality of retrieved parameters.

"Fault symptom" and "fault" of the target machine are determined as a result of the presence / absence of a fault, the fault diagnosis and the simulation of the fault condition. Here, the “fault symptom” is a change in the output status of the target machine (for example, “a copy image is thin” in the case of a copying machine), and the “fault” is a cause of a change in a symbol. This is a change in the mechanism or structure of the target machine (for example, in the case of a copying machine, "a decrease in the light amount of a halogen lamp").

Next, the repair planning section 15 determines, based on the failure diagnosis and the simulation result of the failure state,
A large number of cases stored in the case base storage unit 17 are searched (step S5). Then, a case close to the current state of the target machine is detected, and the order of application is determined (step S6). The detection of this case is based on the failure symptom and whether or not the failures match. Also,
In the determination of the application order, when there are multiple cases that belong to the same class, only the case with the highest priority is applied.

Then, the repair work is executed based on the cases in which the application ranking has been made (step S7). In the repair work, the case is corrected or the repair work is corrected, that is, the work script is refined as necessary, and the corrected case is registered as a new case belonging to the same class. If the case-based repair work is successful, the process ends (YES in step S8), but if the case-based repair work is not successful (N in step S8).
In (O), a repair method is inferred by QMS (step S9), a side effect is simulated (step S10), a repair plan is determined, and repair work is executed based on the determination. (Step S1
1).

Although the inference and the execution of the work in steps S9 to S11 are based on the QMS and not the case in the CBS, if the repair work based on the inference by the QMS is successful, the repair result is obtained. Is registered in the case base storage unit 17 as a new case belonging to another class. Next, a method of failure diagnosis and failure repair will be described in detail with reference to specific examples. In the following description, as an example, a method in which the peripheral portion of the photosensitive drum in a small-sized plain paper copying machine is used as a target machine will be described.

(6-3) Structure and State of Specific Target Machine FIG. 9 is an illustrative view showing a specific target machine. FIG.
In the figure, reference numeral 21 is a photosensitive drum, 22 is a main charging charger, 23 is a halogen lamp for illuminating an original, 24 is a developing device, and 25 is a transfer charger. In this embodiment, for example, three sensors 1a, 1b, 1c are provided. That is, the sensor 1a is an AE sensor for measuring the amount of light that exposes the photosensitive drum 21, the sensor 1b is a surface potential sensor for measuring the surface potential of the photosensitive drum 21,
The sensor 1c is a densitometer for measuring the density of the image copied on the paper.

Although not shown in FIG. 9, three types of actuators are provided. That is, the main charging volume VR1 for changing the main charging voltage of the photosensitive drum 21, the lamp volume AVR for controlling the light amount of the halogen lamp 23, and the transfer voltage between the photosensitive drum 21 and the copy sheet (of the transfer charger 25). Transfer voltage)
Three volumes, namely, a transfer volume VR2 for controlling the control are provided as actuators.

Furthermore, the three sensors 1a, 1 shown in FIG.
The same system control circuit 10 as shown in FIG. 7 is connected to b, 1c and the three actuators. (6-4) Entity model and target model By the way, the target machine shown in FIG. 9 is captured from a physical point of view, and the target machine is expressed as a combination of a plurality of elements at the entity level. If the connection relation between each element is qualitatively expressed using parameters, it is as shown in Table 1. The expression format as shown in Table 1 will be referred to as "entity model".

[0050]

[Table 1]

The expression shown in FIG. 10 in which the entity model is abstracted and expressed as a connection tree of each parameter will be called a "mathematical model". Then, the "entity model" and the "mathematical model" are collectively referred to as a "target model (qualitative model)". The “target model” is qualitative data common to the image forming apparatus, which is also utilized for failure repair described later.

The contents of the physical model and the mathematical model as qualitative data are stored in the target model storage unit 14 (see FIG. 7).
Is stored in Further, in the target model storage unit 14,
For predetermined parameters among the parameters included in the physical model, reference value data measured at the time of factory shipment is stored. This reference value data is
This is characteristic data unique to this image forming apparatus.

[0053] For example, in this machine, as shown in FIG. 11, the parameters X, V _S, O _S, for V _n, respectively, low, normal, reference value data specifying the ranges of high are stored. In this embodiment, the above-mentioned reference value data is responsive to sensing data in the subsequent failure diagnosis and failure repair process, changes in the operating state of the machine, etc.
It is designed to be updated.

In the target model storage unit 14, the evaluation function knowledge as an example of the fault diagnosis knowledge serving as a reference for determining whether or not the target machine is operating normally based on the converted symbols. Is remembered. Note that the evaluation function knowledge, in other words, the failure diagnosis knowledge, may be specific to the target device, may not be specific, and may be common to a wide range of image forming apparatuses.

The evaluation function knowledge of this embodiment includes the following knowledge. Image density O _S = normal, fog degree O _S '<normal, separation performance S _p <normal Here, when O _s , O _S ' and S _p are not the above conditions, the target machine is not operating normally. It will be.

Now, consider the case where the digitized sensor information of the target machine in the normal operation has the following values. Value of AE sensor 1a X = 23 Value of surface potential sensor 1b V _s = 380 Value of densitometer 1c O _s = 7 Also, densitometer 1c when using a blank original document with optical density D = 0
'Determined, the value V _s = dark potential V _n of the surface potential sensor 1b in a state in which off the halogen lamp, and, their values, respectively, head of O _s' value O _s = fog degree O _s = 50 It is assumed that the dark potential V _n = 590.

The fog degree O _s ' and the dark potential V _n may be measured manually, or under certain conditions, for example, every time the power of the target machine is turned on or before the start of copying. Alternatively, it may be programmed to be automatically measured by the sensor.
In this embodiment, the latter is adopted. AE sensor 1
a, the respective values X, V _s , O _s , O _s ′, and V _n obtained by the surface potential sensor 1 b and the densitometer 1 c are respectively
It is converted into a symbol in the digital signal / symbol conversion unit 11. As described above, conversion is performed by each sensor 1
The digital value given from a, 1b, or 1c is compared with the reference value data as the characteristic data stored in the target model storage unit 14, and any one of three types of normal, high, or low is used. Converted to a symbol.

In this embodiment, each parameter is symbolized as follows. X = normal V _s = low O _s = low V _n = low In the fault diagnosis unit 12, these symbolized parameters are function evaluations as an example of fault diagnosis knowledge stored in the target model storage unit 14. Compared to knowledge. As a result, since the image density O _s is not normal, it is determined that there is a failure, and the failure symptom is “the image density is low (O _s
= Low) ”. And then, "O _s
= "Low" as a failure symptom, failure diagnosis, that is, failure inference is performed.

(6-5) Method of Fault Diagnosis The fault diagnosis is first carried out in the fault simulation section 13 by using the mathematical model of FIG. 10 to search for a parameter which may cause O _s = low. . In the mathematical model in FIG. 10, the parameters that may reduce O _s are pointed out, and the results are shown in FIG. In FIG. 12, the parameters marked with an up arrow or a down arrow are the parameters that may cause the parameter O _s = low, and those with an up arrow are those with a down arrow, and those with a down arrow are If that parameter drops, it causes O _s = low.

Next, the O _s searched for in the mathematical model
= Each parameter ζ, D _s , which can cause a low,
With respect to V _t , γ ₀ , V _b , V _s , V _n , X, β, _HL and D, the cause of the parameter change is detected by the failure diagnosis unit 12. This detection is performed based on the substance model of Table 1, and in this embodiment, the following failure candidates are inferred. That is, V _t = low: → defective transfer transformer ζ = low: → deterioration of paper V _b = high: → defective development bias γ ₀ = low: → degradation of toner V _n = low: → defective main charging voltage _HL = High: → Halogen lamp setting error D = Low: → Original is thin The knowledge written on the right side of the "arrow →" in the above failure candidates, that is, defective transfer transformer, paper deterioration, and development bias. The knowledge such as defect, is fault knowledge, and this knowledge is included in the qualitative data common to the image forming apparatus. Among the parameters, β is the sensitivity of the photoconductor, and is not included because it does not increase. D
_s , V _s and X are also excluded because they are represented by other parameters.

Then, in response to the above inference made in the failure diagnosis section 12, the failure simulation section 13 carries out a simulation of the failure state. The failure state simulation is to infer the state of the target machine when the inferred failure occurs, respectively. More specifically, assuming that the cause of O _s = low, that is, the failure is, for example, a defective transfer transformer, V _t = low is set for the model in the normal state.
Then, the influence given to each parameter in that state is examined on a mathematical model.

For example, if V _t = low is set, then O _s
= Low and S _p = low and all other parameters are normal, this is the V obtained from the sensor.
Inconsistent with _s = low and V _n = low. Therefore, we get the result that the reasoning of the failure is wrong. Similarly, ζ = low is set on the mathematical model in the normal state, and the result is compared with the symbol obtained from the sensor. Again, on the mathematical model, V _s = normal, V _n = normal, whereas the symbols from the sensor are V _s = low, V _n =
Since it is low, there is a contradiction, and the reasoning of the failure is judged to be erroneous.

In this way, the failure state simulation is performed for all the failure candidates, and it is confirmed whether or not the inference of the failure is correct. As a result, in the case of this example, when the failure is “defective main charging voltage (V _n = low)”, a result that matches the actual state of the target machine is obtained, and other failure candidates All conclude that they are inconsistent with the actual state of the device.

Therefore, it can be concluded that the failure in this case is a failure of the main charging voltage. Table 2 shows the status of each parameter of the target machine at that time.

[0065]

[Table 2]

FIG. 13 is obtained by tracing the situation of the parameters shown in Table 2 on the mathematical model. In FIG. 13, a downward arrow attached to the right side of each parameter is low, an upward arrow is high, and N is normal. (6-6) Execution of Repair Work Next, the failure diagnosis unit 12 and the failure simulation unit 1
The repair work is executed based on the result of the failure diagnosis performed in 3.

Hereinafter, the repair work will be described step by step according to the flow charts shown in FIGS. (6-7) Failure Repair Processing by Searching for Cases Prior to repair work, failure diagnosis is first performed according to the above-described method (step S21). As a result, it is assumed that the fault symptom occurring is “low image density (O _s = low)” and the fault that causes the fault symptom is diagnosed as “defective main charging voltage”. Table 2 shows the status of each parameter at that time.

Next, based on the result of the above-mentioned failure diagnosis,
A process of determining a case to be applied is performed (step S
22). The specific content of the process shown in step S22 is shown in the flowchart of FIG. With reference to FIG. 17, a case that matches the failure diagnosis result is detected from the cases stored in the case base storage unit 17 (see FIG. 7) (step S221). As described above, the cases stored in the case-based storage unit 17 are hierarchically classified and classified according to the failure symptoms and the failures. Table 3 shows an example of the stored cases.

[0069]

[Table 3]

As shown in Table 3, with respect to the failure symptom of "low image density", either "failure of main charging voltage" or "failure of halogen lamp setting" may be considered. If the main charging voltage is defective, type C
Four cases 001,00 classified into 1, C2, C3
2,004,003 are registered. The three classes C1, C2, and C3 are given priorities, and in this example, the priorities are given in the order of C3, C2, and C1. Further, characteristics are stored in the class C1 to which a plurality of cases belong. The feature of the class C1 is a feature common to a plurality of cases belonging to the class C1. The details will be described later.

If the halogen lamp is not set properly, the type is C
There is only one case 005 that belongs to the category C4. Note that there is no feature in the class to which only one case belongs because no feature can be extracted. Similarly, "image cast"
In the case of such a failure symptom, there are two possible failures: "defective main charging voltage" or "poor halogen lamp setting". Then, in the case of “defective main charging voltage”, the cases can be classified into two classes C5 or C6, and there are three cases belonging to class C5: 006, 008, 009.
Further, the case belonging to class C6 is case 007.
Then, between the classes C5 and C6, the priority of the class C6 is set high. Further, common features of the three cases belonging to the class C5 are extracted and stored.

On the other hand, the failure is "bad halogen lamp setting".
In the case of, only one case 010 belonging to the class C7 is registered. In step S221, cases 001 and 002 belonging to the failure symptoms “low image density” and the failure “defective main charging voltage” obtained by the failure diagnosis are selected from the hierarchically classified cases shown in Table 3. 004, 00
3 is detected.

The contents of these four detected cases are shown in Table 4, Table 5, Table 6 and Table 7. As shown in Tables 4 to 7, in each case, a case number, parameter status before repair, parameter status after repair, failure symptom, failure, repair work number, successful application number and failed application number are stored. Has been done.

[0074]

[Table 4]

[0075]

[Table 5]

[0076]

[Table 6]

[0077]

[Table 7]

Therefore, the parameter statuses of Table 2 representing the current status of the apparatus are compared with the parameter statuses stored in the respective statuses before restoration of Tables 4 to 7 (step S22).
2). As a result of the comparison, in the case 001, all the parameter situations match the parameter situations of Table 2. Case 0
In 02, β is different. In case 003, H _L , X,
β and S _P are different. In case 004, H _L ,
X, V _n and S _P are different.

Therefore, the priority order regarding the case application is determined as 001 → 002 → 003,004.
Next, the case where the degree of coincidence of the parameter situations in the same class is low is deleted (step S223). In this case, the cases 001 and 002 both belong to the class C1, and the case 002 has a lower degree of coincidence in parameter status than the case 001. Therefore, the case 002 is deleted in the application order of the cases. As a result, the order of application is 001 → 003,00
It becomes 4.

Next, it is judged whether or not there are cases in which the degree of coincidence of the parameter states is the same (step S224). In the above case, case 003 and case 004 have the same degree of coincidence in parameter status. Therefore, case 003 and case 004
In applying the and, as to which is applied first, the classes C3 and C to which the cases 003 and 004 belong
The priority of 2 (see Table 3) is taken into consideration (step S22).
5). In this case, since the priority of the class C2 is “2” and the priority of the class C3 is “1”, the case 0 is different from the case 004.
03 is prioritized.

If there is no case where the degree of coincidence of the parameter statuses is the same, the process of step S225 is omitted. Finally, as the application order of the cases, 001 → 003
→ 004 is ranked (step S22)
6). Then, the process for determining the case to be applied is returned.

Instead of the case determination process to be applied described with reference to FIG. 17, the case to be applied may be determined by the process according to the flowchart shown in FIG. In the flowchart shown in FIG. 18, the process of the same step number as that of the flowchart shown in FIG. 17, that is, steps S221 to S224 and S225, S226.
The process of is exactly the same as the description in FIG.

The process shown in the flowchart of FIG. 18 is characterized in that, in step S224, the priority order is determined when there are a plurality of cases in which the degree of agreement between the parameter state before restoration and the current parameter state of the apparatus (Table 2) is the same. As a result, the processing of steps S2241 and S2242 is inserted. Specifically, when there are a plurality of cases in which the degree of coincidence of the parameter statuses is the same (YES in step S224), the number of successful applications and / or the number of failed applications registered in each case are taken into consideration (step S).
2241). In the above example, cases 003 and 0
No. 04 has the same degree of coincidence in parameter status. Therefore,
First, let us look at the number of successful applications of both cases 003 and 004. In case 003, the number of successful applications is “3” as shown in Table 6, and in case 004, the number of successful applications is “1” as shown in Table 7. Therefore, case 003 with many successful applications
Has priority. If the number of successful applications is the same, the case with the least number of failed applications is given priority.

As a result of considering the number of successful applications, or the number of successful applications and the number of unsuccessful applications, if prioritization of application can be determined for a plurality of cases (YES in step S2242), for all detected cases, Priorities for application are assigned (step S226). Even if the number of successful applications and the number of unsuccessful applications are taken into consideration in step S2241, they are equal and the priority order cannot be determined (NO in step S2242).
The priority of the class to which the case belongs is taken into consideration, and the priority is determined (steps S225 and S226).

As described above, in the method of determining the priority order of application based on the consideration of the number of successful applications of the cases or the consideration of the number of successful applications and the number of unsuccessful applications, many repair works have been executed in the past. In this case, the result of the past restoration work will be taken into consideration, and the more the number of past restoration works is, the more beneficial it is. Now, returning to FIG. 14, the description will be continued.

After the case to be applied is decided (step S22), the decided case 001 and the case 001 are decided.
The work script corresponding to is the work script storage unit 1
8 (see FIG. 7 ) and set in the work register (step S23). As mentioned earlier, the case is
They are divided into classes, and cases belonging to the same class share the same work script. In case of Case 001, type C1
And the work scripts shown in Table 8 below are associated with the class C1.

[0087]

[Table 8]

As shown in Table 8, the work script includes
A class C1 that serves as an index is displayed, and a plurality of operations 1,
2, 3, ... Are listed. Each work is described in a rule format and consists of antecedent part status, antecedent part operation and consequent part status. Each work means that the consequent part status is obtained by performing the antecedent part operation in the antecedent part status.

A concrete example will be described with reference to Table 8. For example, in the case of work number 1, the condition of the antecedent part is a condition of parameter H _L = high, and in this condition, the lamp volume AVR is lowered. By operating the antecedent part, the parameter change of parameter H _L = normal, that is, the consequent part situation is obtained.

Since the work scripts are set for each class and the work of the minimum unit is enumerated, there are as many work scripts as there are classes. Referring to FIG.
When the case 001 and the work script of Table 8 are set in the work register (step S23), the status of the parameters before the case 001 set in the work register is set by the repair planning unit 15 (see FIG. 7). (Refer to Table 43) and the condition of the parameter indicating the present condition of the device (Refer to Table 2) are compared to confirm whether or not they completely match (step S24).

Since the parameter status of the case 001 before repair is completely the same as the parameter status of the apparatus,
The process proceeds to step S25, and the work with the number "2" described in the "Repair work" column of the case 001 shown in Table 4 is selected from the work scripts shown in Table 8 and executed. That is, in the antecedent situation _where the parameter V _n = low,
An antecedent operation is performed to raise the main charging volume VR1 (step S25).

As a result of the operation, it is determined whether or not the work is successful, depending on whether or not the consequent part situation described in the work script, that is, the situation where the parameter V _n = normal is obtained (step S26). ). When the work is determined to be successful (YES in step S26), it is further determined whether or not there is the next work (step S).
27). Since only the above work number “2” is described in the “Repair work” column of case 001 shown in Table 4,
It is determined that there is no next work. If the next work number is entered in the "Repair work" column of the case, the process is
Returning to step S25, the work of the work number is selected from the work script and executed, and it is judged whether or not the work has succeeded (step S).
25, S26).

When it is determined in step S27 that the next work is not performed, the numerical value in the "application success number" column in case 001 is incremented by "1", and the success number is updated and registered (step S28). Then, after that, the operation of the priority of the class (1) is performed (step S29). The procedure of the priority operation of the kind (1) performed in step S29 is shown in the flowchart of FIG.

Next, with reference to FIG. 19, the method of the priority operation of the class (1) will be specifically described. By the processing so far, the failure symptom "image density is low",
Case 0 for the device that had the failure "defective main charging voltage"
The repair work was done by applying 01, which was successful. Therefore, the priority of the class to which the successful case 001 belongs is raised (step S291). As shown in Table 3 already described, each case is registered separately for each class, and the priority is set for each class. In the case of the specific example according to this description, the priority of the class C1 to which the case 001 belongs is “3” as shown in Table 3. However, since the repair work based on the case 001 was successful this time, as shown in Table 9, the type C to which the case 001 belongs
The priority of 1 is raised from "3" to "1". That is, the highest priority is given. Also, along with it,
The priority of the other classes C2 and C3 belonging to the failure "defective main charging voltage" with the failure symptom "low image density" is operated. Specifically, the priority of the class C3 is "1" to "2".
In addition, the priority of the class C2 is shifted from "2" to "3" by one. Therefore, 3 which belong to this failure symptom and cause
The priority among the classes C1, C2 and C3 is C1,
The order is changed to C3 and C2.

[0095]

[Table 9]

Next, the class C1 to which the successful case 001 belongs
The characteristic P of is read (step S292). This kind C
The feature P of 1 has the following contents. That is, with respect to the parameter values representing the pre-repair conditions of the plurality of cases 001 and 002 belonging to the class C1, for example, a parameter that is common because it is “not normal” is extracted as a feature of this class.

Two cases 001 and 0 belonging to class C1
The contents of 02 are as shown in Table 4 and Table 5, respectively. Among the parameters listed in the situation before restoration of case 001 shown in Table 4, the non-normal parameters are
V _n, V _s, is a D _s, O _s and S _P. On the other hand, Table 5
Among the parameters listed in the situation before restoration of the case 002 shown in FIG. 3, the non-normal parameters are β, V _n , and V.
_s , D _s , O _s and S _P. Therefore, in these two cases 001 and 002, the parameters that are not normally common are the parameters V _n , V _s , D _s , and O _s.
And S _P , which are extracted and stored as a feature P of that kind. Note that in this case, the parameters to be extracted need not be normal, and a certain parameter is high in case 001 but is not in case 00.
Even if it is low at 2, the parameter is extracted as a characteristic parameter.

Alternatively, among the parameters representing the situation before each restoration, a parameter common to “normal” may be extracted and used as a feature of the class. Note that the features of a class can be extracted only when there are a plurality of cases belonging to the class, and the features cannot be extracted when there is only one case belonging to the class.

In step S292, the characteristic P of the case C1, that is, the parameter V _n in the situation before restoration,
The contents that V _s , D _s , O _s and S _P are not normal are read and stored in a work register. Next, it is determined whether or not there is a class belonging to a failure symptom different from the failure symptom to which the class C1 belongs (step S293). In the case of this specific example, as shown in Table 9, a failure symptom “image fog” different from the failure symptom “image density is low” to which the class C1 belongs
Classes C5, C6 and C7 exist as classes belonging to.

If there are classes belonging to different failure symptoms (YES in step S293), the feature Q of the class is detected (step S294). Specifically, first, type C5
Is detected from Table 9 (this Table 9 is stored in the case-based storage unit 17). Characteristic Q of class C5
Means that the parameters V _n , V _s , D _s and O _s are not normal.

Then, the feature P of the class C1 stored in the work register in step S292 and the feature Q of the class C5 detected in step S294 are compared to determine whether the feature P is similar to the feature Q. (Step S29)
5). The feature P being similar to the feature Q means a case where all the parameters forming the feature P are included in all the parameters forming the feature Q. In this specific example, the parameter forming the feature P includes S _P ,
Since the parameter S _P is not included in the parameters forming the feature Q, the feature P is not similar to the feature Q. Therefore, in the case of this specific example, the process of step S295 is NO.
It is.

In step S295, if the feature P is similar to the feature Q (for example, the parameters forming the feature P are V _n , V _s, and D _s , and the parameters forming the feature Q are V _n , V _s). And D _s or contains other parameters in addition to this)
If so, the process proceeds to step S296. Step S29
In 6, the priority of the class C5 is flagged. This flagging is performed in step S298 described below.
This is added to indicate that the priority should be updated when updating the priority of the class.

Then, it is judged whether or not there is another type to be processed (step S297). The types to be treated include the types C6 and C7 in addition to the type C5.
The processing from step S294 is sequentially performed on 6 and C7. In this specific example, the type C
Since 6 and C7 each belong to only a single case, the characteristics of these classes C6 and C7 cannot be extracted, so that the priorities of these classes C6 and C7 are not flagged.

If there is no other class in step S297, the flow advances to step S298 to set the failure symptom to which the class belongs and the priority of the class included in the failure so that the priority of the flagged class increases. Will be updated. In this specific example, no priority is given to the priority of class C5 (because the process proceeds to NO in step S295).
As shown in Table 9, the priorities of class C5 and class C6 do not change. However, if it is determined in step S296 that the type C5
If the priority is flagged, the step S2
In 98, the priority of the class C5 is set to "1", and accordingly, the priority of the remaining classes C6 is lowered from "1" to "2".

Now, returning to step S26 in FIG.
Even if the work is executed, if the consequent part status described in the work is not obtained, it is determined that the work has failed (step S
No in step 26), the number of failures is updated so that the numerical value in the “application failure number” column of the case (see, for example, Table 4) increases by “1” (step S30). Then, it is determined whether or not there is a case of the next priority (step S31), and if there is a next case (YES in step S31), the process from step S23 is performed on the next case.

If there is no case of the next application priority in step S31, QMS processing is performed (step S32). It is preferable that this QMS process is a process that takes into consideration a secondary effect which will be described in detail later. And Q
It is determined whether the MS processing is successful (step S3).
3) If it is determined that the QMS processing has succeeded (YES in step S33), a new case is created based on the data obtained by the QMS processing (step S3).
4). A new class is created for the created new case, and a case number is added and registered (step S35).

Table 10 shows an example of registration of a new case. Table 1
In 0, the new case 030 is the previous class C1,
It is registered as belonging to a new class C10 different from C2 and C3.

[0108]

[Table 10]

Furthermore, the priority of the class to which the new case 030 belongs is set to "1", and the highest priority is given. Along with this, the priorities of other types belonging to the failure "defective main charging voltage" due to the failure symptom "low image density" are sequentially shifted. That is, as shown in Table 10, the class C1 has a priority of 1 to 2, the class C2 has a priority of 3 to 4, and the class C has a priority of 3 to 4.
The priority of 3 is changed from 2 to 3 (step S3).
6).

If the QMS process is unsuccessful in step S33, no new case is created and the process ends. Now, in step S24 of FIG. 14, when the parameter status before the restoration of the case to be applied does not completely match the parameter status of the failed device, the process proceeds to step S37 of FIG.

Referring to FIG. 15, in step S37,
A case to be applied, this case to be applied is, for example, case 0.
If it is 01, the work of the number “2” described in the “Repair work” column of the case 001 shown in Table 4 is the type C1 shown in Table 8.
Specified from the work script of. Then, the condition V _n = low of the antecedent part of the work 2 is compared with the condition of the parameter V _n of the faulty device shown in Table 2, and it is determined whether or not the conditions of the two parameters match (step S3).
8).

In the case of this specific example, since the parameter status of the failed device matches the status of the antecedent part of the specified work 2, work 2 is executed (step S39). After the execution of the work in step S39, it is determined whether or not the work is successful (step S40), and if it is successful, the presence or absence of the next work is determined (step S41). Step S
In 38, if the parameter V _{n of the} faulty device is normal, V _n, which is the condition of the antecedent of work 2,
= Does not match Lo. In such a case, in order to make the parameter status of the device match the status of the antecedent part of the work, additional processing 1 described in steps S42 to S45 is executed.

That is, the process proceeds to step S42,
In the work script shown in Table 8, it is searched whether or not there is another work that can match the parameter status of the faulty device with the status of the antecedent part of the work 2. That is, in the case of this specific example, it is determined whether or not there is an operation capable of making the parameter value V _n = normal of the faulty device low.

It can be seen from Table 8 that task 5 can change the parameter V _n from normal to low. Therefore, it is determined to be YES in step S43, and the column of "restoration work" of the case 001 shown in Table 4 is changed from "2" to "5, 2".
And the flag A is set to indicate that the temporary correction has been performed (step S44). Next, the work 5 added by the temporary correction is executed (step S4
5) It is determined whether or not the work 5 has succeeded (step S40). When the work 5 is successfully executed (YES in step S40), it is determined whether or not there is the next work (step S41). In this case, since work 2 exists as the next work, the process proceeds to step S37 again, the next work 2 is designated, and the condition of the antecedent part is compared with the parameter condition of the failed device. As a result, the parameter status of the failed device is set to V _n = low by the execution of work 5 in step S45, and therefore matches the status of the antecedent part of work 2. Therefore, in step S38, Y
It is determined to be ES, and the work 2 is executed (step S3
9).

By the way, in step S39 or S45, if the consequent part status is not obtained even after the antecedent part operation is executed, it is determined that the work has failed (step S40). In other words, if the parameter status (the parameter status of the failed device after the work) as a result of performing a certain work does not reach the parameter status (sequel part status) set in the work, the work is regarded as a failure. To be determined.

At this time, the additional process 2 for avoiding the cause of the work failure, which will be described with reference to the flow of FIG. 16 below, is executed. That is, first, all the failure symptoms that are the same as the failure application cases and the cases that belong to the failure are searched, and in the cases, all the work numbers related to the failure are described in the “repair work” column. Is detected (step S51).

[0117] Now, in order to make it easy to understand along with a concrete example, if the application example relating to the failure is 001, Table 3
As shown in FIG. 5, the same failure symptom and the case that belongs to the failure are 002, 003, 004, and both cases have the same work number 2 as the failure number of the case 001 in the “Repair work” column. Work number 2 is described (see Table 4, Table 5, Table 6 and Table 7). Therefore, case 002
~ 004 are detected (step S51).

Then, by comparing the parameter statuses before the restoration of the failure application example 001 and the detected case 002, 003, 004, the common points of the parameter statuses are extracted, and the extracted common points are displayed. The parameter status of the faulty device shown in 2 is compared (step S5).
2). As a result, it is determined whether or not there is a difference between the parameter status before the repair in each case and the parameter status of the failed device (step S53).

Cases 001 to 004 shown in Tables 4 to 7, and
Even comparing with the parameter status of the faulty device shown in Table 2,
In the case of this example no such parameter is found.
No. Therefore, tentatively, before restoration of cases 001 to 004,
V is common in the parameter situation_t= Provisional if low
And the following description will be given. Then,
In step S53, before restoration of cases 001 to 004
A common point in parameter situations, and a faulty device
Parameter V as the difference from the parameter situation _tGatori
Can be lifted. That is, restoration of each case 001 to 004
In the previous parameter situation, commonly V_t= Low
However, in the parameter status of the faulty device, V_t= Norma
It is.

Therefore, in step S53, YES is determined, and it is hypothesized that the parameter V _t = low is the cause of the failure of this work, and the work that can change the parameter V _t from low to normal is The work script shown in FIG. 8 is searched (step S54), and the presence or absence thereof is determined (step S55). Looking at the work script in Table 8, it is found that the work can change the parameter V _t from low to normal, so it is determined that there is work (YES in step S55).

In this case, the "repair work" column of the failure application example 001 (see Table 4) is provisionally corrected, and the work number "3" is inserted prior to the work number "2". Further, the flag B is set to indicate that this temporary correction has been made (step S56). Next, the work 3 is executed (step S57). Then, when V _t = normal as a result of the work 3 being executed, it is determined that the work is successful (YES in step S58).

In this case, V _t = normal is an indispensable condition for the condition of the antecedent part of work 2. Therefore, V _t =
The normal script is corrected to be added, and the work script in Table 8 is rewritten as shown in Table 11 below (step S59).

[0123]

[Table 11]

In the work script shown in Table 11, work 2
The condition of the antecedent part is “V _n = low, (and) V _t = normal”. Then, the processing from step S37 shown in FIG. 15 is performed again. Step S5 of FIG.
If it is determined in step 8 that the work was not successful, it is determined whether or not there is another work in which the parameter V _t can be changed from low to normal in the work script shown in Table 8 (step S60), if there is work, the processing from step S56 is repeated.

When it is determined in step S60 that there is no other work, or when it is determined in step S53 that there is no different parameter, or when it is determined in step S55 that there is no work, or When it is determined that there is no work in step S43 shown in FIG. 15, the process proceeds to step S61, and the flags A and B are reset. Further, in this case, since the case application has failed, the numerical value in the "application failure number" column of the case is increased by "1" (step S62). Then, it is determined whether or not there is another case, that is, a case of the next application priority order (step S63).

If there is a next case (YES in step S63), the case and the corresponding work script are set in the work register (step S6).
4), the processing from step S37 shown in FIG. 15 is performed. On the other hand, when it is determined in step S63 that there is no case of the next application priority order, the process proceeds to step S32 shown in FIG. 14 and the QMS process is performed.

Now, returning to FIG. 15, the process after it is determined in step S41 that there is no next work will be described. In this case, the state of the flag A or B is determined (step S46). When it is determined that neither the flag A nor the flag B is set as a result of the state determination, the parameter status of the faulty device and the status of the antecedent part of the selected work match in the processing up to that point, This is when the work is successful as a result of executing the work. Therefore, in this case, in step S47, the applied case, for example, case 0
The numerical value in the column of "Number of successful applications" of 01 is increased by "1".
Then, the operation of the priority of the class (1) is executed (step S48). This kind of priority operation (No. 1) in step S48 is exactly the same as the operation performed in step S29 of FIG. Therefore, the specific content of the operation of step S48 is omitted here so that the description will not be repeated.

On the other hand, in step S46, the flag A
Consider a case where it is determined that either B or B is set. The case where the flag A is set is the case where the additional processing 1 of steps S42 to S45 is performed. Specifically, the work with the number described in the "Repair work" column of the example was selected from the work script, and the work was attempted, but the parameter status of the failed device is the status of the antecedent part of the work. It is a case where the work cannot be performed because it does not match, and before performing the work, another work for matching the parameter status of the failed device with the status of the antecedent section described above is performed successfully. . In other words, this is a case where the operation of "detailing the work script" described above is introduced, and the detail reference succeeds in the case reference.

On the other hand, the case where the flag B is set is the case where the additional processing 2 of steps S51 to S59 is performed. Specifically, when the work was not successful because the antecedent part status of the selected work was insufficient, the antecedent part status of the work is identified, and
There is work to change the parameter status of the faulty device to match it, and by performing these works, there is no difference between the parameter status of the faulty device and the antecedent condition of the selected case, and the work is successful. . In this case also, in a broad sense, it can be said that the case where the case reference has succeeded due to the detailing has been introduced as a result of introducing the operation "detailing the work script".

Therefore, when the flag A or B is set, the operation of "detailing the work script" is introduced, and the original case is corrected and newly added within the range qualitatively reachable. It means that a case has been generated. Then, the generated new case should be considered to belong to the same class as the original case. This point is as described in the outline of the present invention.

Therefore, in step S46, the flag A is set.
If it is determined that B or B is set, the process proceeds to step S49, the generated new case is registered in the same class as the original case, and then the flags A and / or B are reset. Table 12 shows an example of the stored contents of the case-based storage unit 17 (see FIG. 7) after the new case is registered. As shown in Table 12, case 040 is newly registered as a case belonging to class C1.

[0132]

[Table 12]

Table 13 shows the contents of this new case 040.
Shown in

[0134]

[Table 13]

Comparing the antecedent status of case 040 shown in Table 13 with the antecedent status of case 001 (see Table 4),
It differs from case 001 in that the parameter V _t = low. Therefore, in applying this case 040, it is necessary to change the parameter V _t from low to normal first. Therefore, the work number “3, 2” is described in the “repair work” column of this case 040. Has been done.

After that, a class priority operation (No. 2) is executed (step S50), and the process ends. Next, a specific processing procedure of the type priority operation (No. 2) in step S50 will be described with reference to FIG. 20 and Tables 12 and 13. In the class priority operation (No. 2), first, the generated new case case number 040 is registered in the case C1 to which the currently referred case, for example, the case 001 belongs (step S501). Table 1
As shown in 2, in the column of the case number of the class C1, "001,
In addition to "002", "040" is added.

In the case application of this time, since the restoration process succeeded as a result of applying the new case 040,
The class C1 including this new case 040 has the highest priority. That is, the priority column of the class C1 is rewritten from "2" to "1" (step S50).
2). Further, since the priority of the class C1 is set to "1", the remaining classes C2 and C belonging to the same "fault" as this class
The priorities of C3 and C10 are also sequentially shifted. as a result,
The priority of the class C2 remains “4”, the priority of the class C3 remains “3”, and the priority of the class C10 is shifted from “1” to “2”.

Next, the type C1 in which the new case is additionally registered this time.
Each case (including a new case) belonging to the above, in this specific example, the situation before each repair in three cases of cases 001, 002, 040 is inquired, and the feature P ₁ of this class C1 is extracted (step S503). . How to extract the feature P ₁ is
This is similar to the method of extracting the features described above. As a precaution, a description will be given with reference to a specific example. Case 0 shown in Table 4
01, case 002 shown in Table 5 and case 0 shown in Table 13
Compare the stored parameter status in each of the 40 pre-repair statuses. Then, a parameter that is not normal, that is, a low or high parameter is extracted, and a parameter that is extracted in common with cases 001, 002, and 040 among the extracted parameters is extracted as a parameter that constitutes the feature P ₁ . Specifically, the parameters V _n , V _s , D _s, and O _s are extracted as the parameters forming the feature P ₁ . The parameters forming the feature P ₁ are registered as the features of the class C1 in Table 12.

Next, it is judged whether or not the feature P ₁ extracted this time matches the feature P already registered as the feature of this class C1. As shown in Table 3, the characteristics P of the class C1 registered previously are V _n , V _s , D _s , and O _s.
And S _P , which does not match the feature P ₁ extracted this time in that S _P is included. Therefore, in the case of this specific example, NO is determined in step S504.

Then, in this case, as shown in Table 12, the characteristic of the class C1 is rewritten to the characteristic P ₁ extracted this time. That is, the feature P ₁ extracted this time is re-registered as the feature of class C1 (step S506). Next, it is determined whether or not there is a class belonging to a failure symptom different from the failure symptom “low image density” to which the class C1 belongs (step S50
6). In the case of this specific example, as shown in Table 12, as a class belonging to different failure symptoms, the failure symptom "image fog"
There are classes C5, C6 and C7 belonging to Therefore, in the specific example, YES is determined in step S506.

In this case, the features of the classes C5, C6, C7 and the feature P _{1 of the} class C1 described above are sequentially compared. Specifically, first, the feature Q of the class C5 is detected (step S507). As shown in Table 12, the characteristic Q is that the parameters V _n , V _s , D _s, and O _s are not normal.

Next, the feature P _{1 of the} class C1 and the feature Q of the class C5 are compared, and it is determined whether or not the feature P is similar to the feature Q (step S508). In this example, as shown in Table 12, the feature P ₁ has parameters V _n , V _s ,
It is composed of D _s and O _s , and the feature Q is composed of parameters V _n , V _s , D _s, and O _s . Therefore, P ₁ = Q. Therefore, the feature P ₁ is similar to the feature Q. Therefore, the priority of the class C5 is flagged (step S509).

Then, it is judged whether or not there is another type to be processed (step S510). Since the types to be processed include C6 and C7 in addition to the type C5, these are sequentially processed in accordance with steps S507 to S510. It should be noted that both C1 and C7 are 1
Since only one case belongs to these classes, there are no characteristics of these classes, so the priority levels of these classes C6 and C7 are not flagged.

Then, in step S511, a fault "defective main charging voltage" including the flagged class C5.
The priorities of the two classes C5 and C6 belonging to are updated. Specifically, the priority of the class C5 having the feature Q similar to the feature P ₁ is given the highest priority, and the priority of the class C5 is changed from “2” to “1”. Along with this, the priority of the class C6 is changed from "1" to "2".

As described above, the priority of the class C1 to which the case successfully applied this time belongs is raised, and in the class belonging to a failure symptom different from the failure symptom to which the case belongs,
The priority will be increased for classes with similar characteristics to the cases successfully applied this time. As a result, it is possible to improve the case retrieval efficiency by using the “class”. because,
Since the classification of cases by "class" is beyond the framework of the classification of cases by "fault symptom" and "failure", in other words, the case classification by QMS processing, it is classified by another classification method that exceeds the framework. With respect to the selected cases, the priority order for applying the cases is increased for those having common characteristics. By doing so, it is possible to efficiently deal with the phenomenon that the root cause is often the same in an actual mechanical system, but the manifested fault symptoms are different, although the root cause is different.

Further, in the processing for determining the case to be applied described with reference to FIG. 17, the device parameters and the parameters before the case restoration are compared in step S222, and as a result, the same type is determined in step S223. The cases in which the degree of agreement of the parameter status within the group was low were deleted. FIG.
The same applies to the process of deciding the case to be applied as shown in FIG.

However, the process of determining the case to be applied is not limited to this, and the process of deleting the case in which the degree of coincidence of the parameter status in the same class is low may be omitted in step S223. . That is, in the process of determining the case to be applied illustrated in FIG. 17 or 18, the process of step S223 may be omitted. Step S2
When the process of 23 is omitted, the case to be applied may include a plurality of cases belonging to the same class. Therefore, in this case, each time the previously applied case fails, it is only necessary to determine whether or not the next applied case belongs to the same class as the previously applied failed case. Specifically, FIG.
The process after step S30 shown in FIG. 4 may be slightly modified, and the process after step S62 shown in FIG. 16 may be slightly modified.

21 (A) and 21 (B) show a concrete example of the processing for making a correction. Next, with reference to FIG. 21, after the failure count is updated in step S30, the class to which the failure-related case belongs is stored in the buffer memory or the like (step S100). And then
It is determined whether or not there is a case with the next priority (step S3).
1). If there is a next case (step S31
Then, it is determined whether or not the next case belongs to the same class as the case related to the failure (step S101). If the next case does not belong to the same class as the failure case, the process proceeds to step S23 (FIG. 14). If the next case belongs to the same class as the failure case, the process returns to step S31, and the case It is determined whether or not there is a case with the next priority of. In this way, immediately before applying the case, it is determined whether or not the case to be applied belongs to the same class as the failure case, and if it belongs to the same class, the application of the case is omitted. Similarly, referring to FIG. 21B, step S62
After the failure number increase processing is performed in step S1, the class to which the failure case belongs is stored in the buffer memory or the like (step S102), and it is determined whether or not there is another case, that is, the next application priority order case. If there is a next case (YES in step S63), it is determined whether or not the case having the next application priority order belongs to the same class as the class to which the failure-related case stored in step S102 belongs. Is performed (step S103). If the case with the next application priority order does not belong to the same class as the class to which the failure occurs, the process proceeds to step S64 (FIG. 16). However, if the case of the next application priority order belongs to the same class as the case to which the failure-related case belongs, the process returns to step S63,
It is determined whether or not there is a case of the application priority next to the case. As a result, immediately before applying the case, it is determined whether the case is the same class as the class to which the failure occurs, and if the case is the same class, the application is omitted.

When the restoration work is executed, it is particularly effective to use the above-described technique of case retrieval and case application for a small ordinary copying machine or the like as described in the above concrete examples. This is because an apparatus represented by a small-sized ordinary copying machine has an unstable element (for example, positively utilizing chemical changes) as a control target in its constituent system. Therefore, the relationship between the sensor parameter and the actuator parameter may change due to changes in the state in which the constituent system is placed, such as environmental changes and structural deterioration. The case search in the description of the above specific example can be said to be because the device collects the changes between the parameters during running, causes a kind of learning using the changes, and tunes the knowledge. Even if there is a change between the parameters, the repair work can be effectively performed. That is, when the relationship between the parameters of the target machine changes, the case is modified based on the change to create a new case, and the content of the work script is modified.

(6-8) Inference of Repair Plan Next, the method of the QMS process shown in step S32 of FIG. 14 and the inference of the side effect in the process will be described. As a result of the failure determination, “image density is low (O _s = low)” is taken as a failure symptom, and the goal of repair is to increase O _s .

Therefore, the relationship on the mathematical model shown in FIG.
From D_sOr raise V_tOr raise
Depends on whether ζ is increased or_sTo
It is inferred that it can be raised. Then D_sOn
If you make an inference with the goal of raising_sRaise
Or V _bOr γ₀To raise
Get one of the conclusions. Thus, based on the mathematical model
Based on the repeated inference,
Can be obtained on a mathematical model. The results obtained are
As shown in Table 14.

[0152]

[Table 14]

By the way, repair candidates obtained based on the mathematical model include those that can be realized and those that cannot be realized. For example, D: the optical density of the original cannot be changed, and β: the sensitivity of the photoconductor is difficult to change.

Γ ₀ : Toner sensitivity cannot be changed, and ζ: Paper sensitivity cannot be changed. Further, in this specific example, V _b : bias voltage cannot be changed because there is no actuator. Of course, V _b can be made variable by adding an actuator.

Further, X: logarithm of the reflected light amount of the document V _s : surface potential of the drum after exposure D _s : toner density on the drum, it is impossible to change itself, and other parameters can be indirectly set. It can only be changed by changing it, and is excluded from the repair candidates here.

Although not directly related to this specific example, the amplitude of _Asp : AC voltage for separation can also be changed by adding an actuator. Depending described above, in this embodiment, as the repair candidates, V _t: transfer voltage V _n: surface potential after principal charge H _L: logarithm of halogen lamp output quantity of light is taken up.

On the other hand, the following knowledge is stored in advance in the target model storage unit 14 as the repair plan knowledge. That is, (a) increase V _t → increase transfer transformer control voltage (b) decrease V _t → decrease transfer transformer control voltage (c) increase V _n → increase main charging transformer control voltage Raise (d) Decrease V _n → Decrease control voltage of main charging transformer (e) Raise _HL → Shift halogen lamp control signal to high voltage side (f) Decrease _HL → Halogen lamp control Shift the signal to the low voltage side. It is. The repair plan knowledge stored in the target model storage unit 14 is characteristic data unique to this device. By applying the repair plan knowledge to repair candidates obtained based on a mathematical model, as a repair operation for increasing O _s , (a) increase V _t → increase the control voltage of the transcription transformer (c ) There are three methods of increasing V _n → increasing the control voltage of the charging transformer (f) decreasing H _L → shifting the halogen lamp control signal to the low voltage side.

Even if any one of these three methods is executed as long as the image density O _S is simply increased,
It can be repaired. However, the target machine, by increasing the image density O _S, it is conceivable to undergo various side effects. Therefore, in this embodiment, as described below, inference of the secondary effect is performed based on a mathematical model.

(6-9) Inference of secondary influence When three restoration plans derived in the inference of the restoration plan are developed on the mathematical model, FIGS. 22 to 27 are obtained. That is, when (a) V _t is increased, FIG. 22 and FIG. 23 (O _s ' when D = 0 is shown in the mathematical model in FIG. 23), (c) V _n 24 and 25 (in FIG. 25, O _s ' when D = 0 is represented on the mathematical model), (f) H
26 and 27 (FIG. 27 shows D when _L is lowered)
O _s ' when = 0 is represented on the mathematical model).

Then, if the function is evaluated based on the mathematical model, the following state is inferred. That is, (1) When V _t is increased (FIGS. 22 and 23) (a) The output image density is increased. (B) When D = 0, O _s ′> normal.
That is, fogging may occur.

(C) S _p > normal, and there is a possibility that separation failure will occur. (2) When V _n is increased (FIGS. 24 and 25) (a) The output image density is increased. (B) When D = 0, O _s ′> normal, and fog may occur. (3) When H _L is lowered (FIGS. 26 and 27) (a) The output image density only increases, and there are no other secondary effects.

Therefore, the repair planning unit 15 selects a repair plan having the least secondary effect, that is, a repair plan in which H _L is lowered. This repair plan is consistent with the operation for eliminating the failure obtained by the failure diagnosis.
In other words, from a different point of view, the reason for failure in failure diagnosis is to trace the actual state when the device fails on a mathematical model and grasp the state of each component when the device fails to While estimating
In the estimation of the repair plan, the condition of the device is traced on a mathematical model on the assumption that the device is in the normal state, not the failure state, and the repair plan is inferred based on the trace. In the above-described specific example, the same failure and repair plan are obtained as a result of inference in failure diagnosis and inference in repair plan.

However, in some cases, the reason for the failure diagnosis is based on the device in the failure state, whereas the reason for the repair plan is based on the device in the normal state. Therefore, the results obtained by the two differ. Sometimes. In such a case, if only those that do not contradict the conclusion obtained in the inference process of failure diagnosis are selected in inferring the repair plan,
Inference processing of repair plans can be performed in a shorter time.

In the above case, if the restoration plan of lowering H _L cannot be selected, for example, if the AVR volume for shifting the halogen lamp control signal to the low voltage side is already at the lower limit, A repair plan is selected that raises V _{n in} (2), which has less side effects. However, when a repair plan of increasing V _n is selected, a side effect of possibility of fogging is predicted, and therefore FIG.
In the mathematical model of FIG. 25, which parameter should be manipulated to lower O _s ' is examined based on the mathematical model of FIG. 25, and the manipulation is selected based on the repair plan knowledge. As a result, either H _L is increased, V _n is decreased, or V _t is decreased, and a repair plan including prevention of fogging is performed.

That is, as shown in FIG. 28, the inference of the repair operation is developed assuming the side effect. In the inference expansion of the repair operation as shown in FIG. 28, (a) the branch that is inconsistent with the previous repair plan on the mathematical model is not selected (b) the one with the least side effect is selected (c) The expansion is performed based on the knowledge that the one that forms the loop stops expanding at that point.

In FIG. 28, after all, there are two repair plans of (1) a loop of V _n ↑ → H _L ↑ → V _n ↑ and (2) a loop of V _n ↑ → V _t ↓ → V _n ↑. Remain. Now, suppose that when the loop of (1) is performed as a repair plan, the image density becomes appropriate, that is, O _s becomes normal. In such a case, since V _n and H _L are increased, the image density O _s is increased.
When the sensor 1b returns to the normal state before restoration
The value of the surface potential measured by should have changed significantly higher than the value initially measured.
However, since the repair work has succeeded, the state of the parameter V _s after repair must be normally symbolized. Therefore, in such a case, when the restoration is completed, the reference data for symbolizing the parameter V _s shown in FIG. 11 is changed based on the measurement value measured by the sensor 1b, and is rewritten to the data shown in FIG. 29, for example. . In this way, the reference data is updated as needed after the repair work is completed.

In this embodiment, when the above-mentioned loop (1) in FIG. 28 is carried out, specifically, the main charging volume VR1 is operated to raise the surface potential of the photosensitive drum 21, thereby obtaining the gain. When a rash occurs on the copy to be made, the lamp volume AVR is operated to increase the light amount of the halogen lamp, and the image density of the copy is reduced.

Then, while the main charging volume VR1 and the lamp volume AVR are alternately increased appropriately, when the image density becomes normal, that is, when the parameter O _s becomes normal, the detection output of the densitometer which is the sensor 1c. , The repair process is terminated. Further, when the above two repair plans are infeasible, the above repair plan of increasing V _{t in} (3) is selected, and the secondary effects of fogging and separation failure are two. The fault diagnosis is performed assuming one of the two, and a repair plan is selected.

Then, the selected repair plan is carried out, and in the case of loop processing, it is judged as a failure when the operation of the parameters on the loop reaches the limit. Further, in the case of this embodiment, as described in the specific example, the end of repair is determined when O _s becomes normal, and the repair is stopped in that state. In the inference of the secondary effects described above, the three repair plans derived in the inference of the repair plan are sequentially developed on a mathematical model, and the secondary effects are inferred together for each of the three repair plans. ing. Instead of inferring such side effects, the following processing may be performed.

That is, it is assumed that, for example, three repair plans are derived in the inference of the repair plans. In that case,
Taking only one of the three repair plans, simulating the side effects that may occur if the actuator means is actuated based on the repair plan, the simulated side effects are , It is determined whether or not the removal can be performed by operating the actuator means other than the actuator means selected by the repair plan.
Then, when it is determined that the secondary effect can be removed, the actuator selected by the repair plan is actually operated, the repair is executed, and the secondary effect is operated by activating the other actuator means. Remove it.

As a result, the other two derived from the restoration plan
There is no need to simulate side effects based on one plan, and overall repair operation time can be shortened. In the above case, if the first selected repair plan simulates a side effect, it is determined that the simulated side effect cannot be eliminated by actuating another actuator means. If so, the first repair plan may be abandoned, then the second repair plan may be taken up, and the actuator means selected based on the second repair plan may be activated. Simulates no side effects, and determines whether the simulated side effects can be removed by actuating an actuator means other than that actuator means, and removes the side effects. When I can,
Repair work is performed based on the second repair plan.

In this way, of a plurality of repair plans derived in the inference of the repair plan, a certain first repair plan is extracted, the side effect in that case is inferred, and the side effect is extracted. If it can be removed, the repair based on the first repair plan is immediately executed. If the remediation plan has too much side effects, it is abandoned, the next remediation plan is selected, and the side effects in that case are simulated.

In such a case, regarding which repair plan is selected first out of the plurality of repair plans derived in the inference of the repair plan, for example, the failure cause obtained in the failure diagnosis is selected in consideration of the failure cause. Is preferred. In the above embodiments, since the number of parameters of the actuator is small, the repair itself is considerably limited. However, the flexibility and possibility of the repair can be further improved by increasing the number of parameters of the actuator.

In the specific example described above, if any of the repair work is successful, it is determined that the state of the device after the success is a normal state. It is preferable that the reference value data (reference value shown in FIG. 11) of each parameter is updated by the data value, and the parameter is symbolized based on the new reference value data.

(6-10) Others Further, in the above-mentioned specific example, although the operating range of each actuator is not particularly mentioned, the target model storage unit 14
If the operating range data that sets the operating range of the actuator is included in the device-specific feature data stored in the actuator, the actuator can be operated when the output state of the actuator is within the stored operating range. When the output state of the actuator reaches the upper limit or the lower limit of the stored operating range, it is determined that the operation of the actuator is not possible, and can be used for determining whether the repair work is correct or not.

Further, in the above-described specific example, the system for automatically performing self-diagnosis and self-restoring based on the change in the sensor output is taken up. However, the image forming apparatus has a self-diagnosis mode setting key, etc. May be provided, and self-diagnosis and / or self-repair may be performed only when the self-diagnosis mode setting key is operated.

[0177]

Further, according to the present invention, when inferring the repair plan, the actuator adjustment range may be taken into consideration and only the actuator that is actually adjustable may be selected. More specifically, when the actuator is, for example, an AVR, the lower limit value of the AVR is “0”, the upper limit value is “100”, and the setting state of the AVR can be detected by any integer value from 0 to 100. Configure Further, the lower limit value “0” and the upper limit value “100” of AVR are set in the target model storage unit 14. Therefore, AV
When R is adjusted to a certain state, the adjustment state of AVR is grasped as any integer value data of 0 to 100 corresponding to the adjustment state.

In the repair planning unit 15, whether the AVR adjustment state can be grasped by the integer value data of 0 to 100 obtained according to the AVR adjustment state, and whether the AVR can be selected as the actuator for failure repair is determined. Determine whether.
That is, the lower limit value and the upper limit value of AVR stored in the target model storage unit 14 are compared with the current adjustment state value, and A
It is further determined whether or not VR can be operated in the lower limit direction or the upper limit direction.

Therefore, by adopting such a configuration for each of the plurality of actuators or for any actuator therein, the inference result of the repair plan is the actuator means that can actually operate. It is output as a combination, and there is an advantage that a practical restoration plan can be inferred. The above description is an example of how to set the operation range, and the operation range may be set by another method and compared with the actual actuator state.

Further, not only is the repair planning unit 15 comparing the set adjustable range of the actuator with the actual adjustment value, but also the failure diagnosis unit 12 performs the failure diagnosis of the set actuator when the failure diagnosis is performed. The adjustable range and the actual adjustment value may be compared and referred to. Furthermore, in the image forming apparatus according to the embodiment of the present invention, for example, a manually operated self-diagnosis mode setting key or switch is provided as the self-diagnosis mode setting means, and the self-diagnosis mode setting key or switch is operated. The self-diagnosis or the self-diagnosis and the self-repair described above can be performed only when the above is performed.

The self-diagnosis mode setting key or switch may be arranged at any position, but it is preferably at a position different from the operation keys for normal image formation, for example, the front surface provided in the image forming apparatus. It may be better to install it inside the device that can be operated with the panel open.

[0183]

According to the present invention, it is determined whether or not a failure has occurred in the image forming apparatus.
The failure symptom, the failure cause, and the state of the device are inferred. Then, based on the inference result, a plurality of prestored cases are searched, a case most suitable for failure repair is detected, and failure repair processing based on the case is performed.

In the failure repair processing according to the present invention, first, a case corresponding to a failure is selected from the cases stored by classifying, and when there is a case with the same degree of correspondence, the application priority is the highest. The same kind of case is applied to the failure repair, and the priority is updated according to the result. It has been empirically known that in the actual mechanical system, the next failure often occurs in association with the previous failure. Therefore, by updating the priority regarding case application as described above,
The appropriate case can continue to be selected for possible failures.

If failure repair is not successful depending on the case application, repair work is inferred by so-called qualitative inference using qualitative data, and failure repair is performed based on the inferred work. Then, when the failure is successfully repaired by the inferred work, the inferred work is registered as a case belonging to a new class different from the classes up to then. Therefore, when a new case is created by qualitative inference, the case is registered as a case belonging to a new class,
It is possible to enrich a variety of types, and it is possible to more appropriately deal with a failure that occurs thereafter.

Further, according to the present invention, the cause of the failure is based on the qualitative data common to the image forming apparatus. Therefore, the self-diagnosis repair system capable of handling an unknown failure not explicitly described. The image forming apparatus having the above can be provided. Furthermore, the self-diagnosis restoration system according to the present invention can be commonly applied not to a specific image forming apparatus but to many types of image forming apparatuses, and as a result, an inexpensive self-diagnosis system can be used. An image forming apparatus having a diagnosis and self-repair system can be provided.

[Brief description of the drawings]

FIG. 1 “Detailed work script” in the present invention
FIG.

FIG. 2 is an illustrative view for explaining a method of classifying cases in a case-based repair planning system according to a prior invention.

FIG. 3 is an illustrative view for explaining a method of classifying cases in the case-based repair planning system according to the present invention.

FIG. 4 is a diagram for explaining how to classify when the concept of “class” is introduced into the classification of cases.

FIG. 5 is an illustrative view showing a relationship between classified cases and work scripts.

FIG. 6 is an illustrative view for explaining a requirement for the case set A belonging to different failure symptoms to be similar to the case set B.

FIG. 7 is a block diagram showing a system configuration of an embodiment of the present invention.

8 is a flowchart showing an outline of a processing operation of the system control circuit shown in FIG.

FIG. 9 is a diagram illustrating a schematic configuration of a small-sized plain paper copying machine according to an embodiment of the present invention and three sensors provided in the copying machine.

FIG. 10 is a diagram showing a mathematical model of a copying machine according to an embodiment of the present invention.

FIG. 11 is a diagram showing reference value data of each parameter required when symbolizing each parameter in the copying machine according to the embodiment of the present invention.

FIG. 12 is a diagram showing development for failure diagnosis on the above-described mathematical model.

FIG. 13 is a diagram showing development for failure diagnosis on the above-described mathematical model.

FIG. 14 is a flowchart showing a part of a repair work process to which a case is applied in one embodiment of the present invention.

FIG. 15 is a flowchart showing a part of a repair work process to which a case is applied in one embodiment of the present invention.

FIG. 16 is a flowchart showing a part of a repair work process to which the case is applied in one embodiment of the present invention.

FIG. 17 is a flowchart showing an example of “processing for determining a case” to be applied in the embodiment of the present invention.

FIG. 18 is a flowchart showing another example of the “process for determining a case” to be applied in the embodiment of the present invention.

FIG. 19 is a flowchart showing a process of “class priority operation (No. 1)” in one embodiment of the present invention.

FIG. 20 is a flowchart showing a process of “class priority operation (No. 2)” in one embodiment of the present invention.

FIG. 21 is a partial flowchart for explaining another control for making only one case applied to each class in one embodiment of the present invention.

FIG. 22 is a diagram showing a development for a side effect inference on the above-described mathematical model.

FIG. 23 is a diagram showing a development for secondary influence inference on the above-mentioned mathematical model.

FIG. 24 is a diagram showing a development for secondary influence inference on the above-described mathematical model.

FIG. 25 is a diagram showing a development for secondary influence inference on the above-mentioned mathematical model.

FIG. 26 is a diagram showing a development for secondary influence inference on the above-mentioned mathematical model.

FIG. 27 is a diagram showing a development for secondary influence inference on the above-mentioned mathematical model.

FIG. 28 is a diagram showing an operation for selecting a repair plan in the embodiment of the present invention.

FIG. 29 is a diagram showing the updated reference value data when the reference value data shown in FIG. 11 is updated as a result of the operation of selecting the repair plan in the embodiment of the present invention.

[Explanation of symbols]

 1a, 1b, 1c Sensor 6a, 6b, 6c Actuator 10 System control circuit 11 Digital signal / symbol converter 12 Failure diagnosis section 13 Failure simulation section 14 Target model storage section 15 Repair planning section 16 Symbol / Digital signal conversion section 17 Case base Storage 18 Work script storage

Continuation of the front page (56) Reference JP-A-58-66967 (JP, A) JP-A-58-94012 (JP, A) JP-A-58-221856 (JP, A) JP-A-62-35916 (JP , A) JP 63-233655 (JP, A) JP 3-27058 (JP, A) JP 3-7963 (JP, A) JP 62-52601 (JP, A) JP 1-219697 (JP, A) JP-A 1-169611 (JP, A) JP-A 62-23328 (JP, A) JP-A 1-278865 (JP, A) JP-A 1-291918 (JP, A) JP-A-2-235074 (JP, A) JP-A-2-302828 (JP, A) JP-A-4-74224 (JP, A) JP-A-2-113262 (JP, A) JP-A-63 -70268 (JP, A) JP-A-4-130330 (JP, A)

Claims (1)

(57) [Claims] 1. A case storage unit for storing cases in which work for repairing a failure of an image forming apparatus is described by class, and a priority storage unit for storing application priority for failure repair of each class. When a failure occurs in the image forming apparatus, a case corresponding to the failure is selected, and when there is a case with the same degree of correspondence, the case is selected in order of the highest application priority, and means for applying to the failure repair is selected. And means for updating the application priority of the class to which the case belongs stored in the priority storage means according to the application result of the case, and the image forming apparatus when failure repair by application of the case is not successful. Using the qualitative data predetermined for
When the failure is successfully repaired by the work inference means that infers the work for repairing the failure and the work inferred by the work inference means, a new kind of framework is created in the case storage means, and the inferred work is added to the new kind. An image forming apparatus, comprising: a registration unit that registers as a case belonging to, and a unit that maximizes the application priority of the new class and updates the application priority of other classes accordingly.