JP4887878B2 - Failure diagnosis system and failure diagnosis method - Google Patents

Description

  The present invention relates to a failure diagnosis system and a failure diagnosis method for an image forming apparatus, and more particularly, to a failure diagnosis system and a failure that can uniformly and quantitatively determine the type of defect included in an output image output from an image forming apparatus. It relates to a diagnostic method.

As a result of the complexity of failure modes due to the multi-functionality, high functionality, and high performance of image forming apparatuses, it has become difficult for even an expert to identify the cause of the failure. Therefore, there is a need for a failure diagnosis system that supports the identification of the cause of the failure of the image forming apparatus. As a failure diagnosis system with such a function, the feature value of the defect is extracted by comparing the image with the defect and the reference image without the defect, and the cause of the failure is identified based on the extracted feature value. The technique to do is known (for example, refer patent document 1).
JP 07-92103 A

  By the way, in the failure diagnosis apparatus as described above, since the defect type and the characteristic value to be acquired are directly related, in order to specify the defect type, the defect type is qualitatively estimated and estimated. It was necessary to individually acquire defect feature values specific to the defect type. Therefore, since it is difficult to estimate the defect type in an image forming apparatus having various types of defects, the accuracy of determining the defect type is greatly influenced by the estimation ability of the operator. In addition, since the types of defects are various, the defect feature values are also various, and the labor of the operator who individually acquires the defect feature values is excessive.

  The present invention has been made in view of the above problems, and the object of the present invention is to provide a failure diagnosis system that can determine a defect type quantitatively and with high accuracy based on a defect feature value obtained uniformly. It is to provide a failure diagnosis method.

The fault diagnosis system according to the present invention specifies a defect type set whose element is a defect type which is a type of defect included in defect image information, from defect image information which is information acquired from an image having defects output by itself. based of the first calculating means for calculating a current Gotoku symptom value, to the collector Gotoku symptoms value calculated by the central characteristic value and the first calculation means is calculated in advance for each defect type set as the center of the defect type group The distance between the center feature value and the set feature value is calculated for each defect type set, and the defect type set corresponding to the center feature value used for the smallest of the calculated distances is specified . specifying means, from the defect image information, a second calculating means for calculating a seed-specific feature values for specifying the defect type on the basis of the species-specific feature values calculated by the second calculating means, first It identified defect type by specifying means The individual specific conditions determination processing respectively corresponding to the case, is characterized in that it comprises a second specifying means for specifying a defect type having defective image information.
According to this configuration, it is possible to identify to quantitatively and uniform defect type group by using a condensing Gotoku symptoms value. Thus, for example, the operator guesses a defect in the test pattern, and qualitatively or individually obtains and analyzes the value that characterizes the set for each defect type set having the estimated defect type as an element. Compared to the identification method, not only can the operator's labor be reduced, but the accuracy of identifying the defect type set is not affected by the superiority or inferiority of the operator's ability to infer the defect type set.
In addition, according to this configuration, in order to specify the defect type individually and specifically after the defect type set is uniformly specified by the cluster analysis, for example, a method of uniformly specifying the defect type only by the cluster analysis and Compared with this, it is possible to improve the accuracy of specifying the defect type.

In the above structure, first calculation means, from the defect image information in which each pixel a defect image information is represented in a multi-level, to calculate the current Gotoku symptom value, second calculating means, a defect image information each pixel Te from the defect image information expressed in binary, to calculate the species-specific feature values, can adopt a configuration.
According to this configuration, the feature value calculated from the test pattern expressed in multiple values strongly expresses the feature related to the print density of the defect type as compared to the feature value calculated from one expressed in binary. It does not show the characteristics of the defect type shape very strongly. In addition, the feature value calculated from the test pattern expressed in binary does not express the feature related to the print density of the defect type too strongly compared to the feature value calculated from expressed in multiple values. It strongly expresses the characteristics related to the shape of the type. As a result, since the defect type can be specified from the print density and shape of the defect type, for example, the accuracy of specifying the defect type is improved as compared with the case of specifying the defect type from only multiple values or only binary values. be able to.

In the above configuration, further comprising a concealment means for concealing defect type identified from the defect image information by the second specifying means, first calculation means, from the defect image information concealed by concealing means collecting Gotoku symptoms value recalculated, first specifying means, re-identifying the defect type set based on the current Gotoku symptoms value recalculated by the first calculation means, second calculation means, the defect image information concealed by concealing means recalculated species-specific feature values from the second specifying means, based on the recalculated species-specific feature values by the re-specified defect type group and the second calculating means by the first specifying means, defects A configuration that re- specifies the type can be adopted.
According to this configuration, when the test pattern has a plurality of defects, and the plurality of defects are configured from a plurality of types of defect types, the defect types already identified as the defect types constituting the defects are concealed. In order to re- specify , a plurality of types of defects constituting the defect can be specified .

In the above structure, the second specifying means, re-identifying the defect type only if the length of the current Gotoku Chochi vector with the current Gotoku symptoms value recalculated by the first calculating means element exceeds a predetermined threshold value The configuration can be adopted.
According to this configuration, since the length is the defect of the current Gotoku symptom value is close to the value 0 if not present, the length of the current Gotoku Chochi vector, whether the test pattern is defective Can be determined quantitatively and uniformly. Therefore, for example, it is possible not only to reduce the operator's labor as compared with the case where the operator determines the presence / absence of a defect qualitatively or individually, but also the defect type set discrimination accuracy estimates the operator's defect type set. Unaffected by superiority or inferior ability.

In the above configuration, failure diagnosis means for diagnosing a failure of each component constituting the image forming apparatus by analyzing a failure diagnosis model that models the cause of the failure of the image forming apparatus, and input to the failure diagnosis model An internal state information acquisition unit that acquires internal state information of the device, and a feature amount extraction unit that extracts a feature amount that characterizes the defect with respect to the defect type specified by the second specifying unit, and the failure diagnosis unit includes Further, it is possible to adopt a configuration in which the cause of the failure is identified by analyzing the failure diagnosis model corresponding to the defect type identified by the second identifying means using the information on the feature amount and the internal state information.
According to this configuration, the feature amount extraction and the acquisition of the internal state information are performed without the user, so that it is possible to improve the efficiency by eliminating the trouble of inputting the defect information one by one by the user, and there is no specialized knowledge about the device. However, detailed and accurate diagnosis is possible. In this specification, a user refers to a person who performs an input operation of input information for performing a failure diagnosis.

In the above configuration, current Gotoku Chochi the integral value of the gradation value histogram, tone value projecting the sum of the waveform width gradation value exceeds the threshold value in the waveform high gradation value waveform width, the tone value histograms The standard deviation and the standard deviation of the gradation value projection waveform, the number of peaks of the gradation value projection waveform, the maximum value of the gradation value projection waveform, and the intensity of the frequency component of the gradation value projection waveform can be adopted.
According to this configuration, the size of the defect area is increased by a certain width or more in the scanning direction (or sub-operation direction) due to the number of high gradation value pixels of the gradation value projection waveform due to the integral value of the gradation value histogram. The width of the sub-scanning direction (or the operation direction) varies depending on the standard deviation of the gradation value histogram and the standard deviation of the gradation value projection waveform. The number of defects of the defect set is such that the magnitude of the defect gradation value is based on the maximum value of the gradation value projection waveform, and the presence or absence of the periodic occurrence of the defect is based on the intensity of the frequency component of the gradation value projection waveform. Feature values that quantitatively represent features can be obtained uniformly. Therefore, “line / vertical line” that is a defect type set having “point” and “vertical line” as defect elements, and a defect type set that has “point” and “horizontal line” that are defect types as elements. Only “line / horizontal line”, defect type “cover”, “horizontal line” and “vertical line” as defect element set “vertical line / horizontal line / cover”, defect type “vertical line” Quantify which set of defects the test pattern belongs to, the “vertical line” that is the defect type set as the element, or the “horizontal line” that is the defect type set that has only the defect type “horizontal line” as the element And can be determined uniformly. Thus, for example, the operator guesses a defect in the test pattern, and qualitatively or individually obtains and analyzes the value that characterizes the set for each defect type set having the estimated defect type as an element. Compared to the identification method, not only can the operator's labor be reduced, but the accuracy of identifying the defect type set is not affected by the superiority or inferiority of the operator's ability to infer the defect type set.

In the above configuration, by JP Chochi species, the maximum value of the projection waveform, and the standard deviation of the spectral intensity of the maximum value and the projection waveform spectral intensity in the specific frequency region of the projection waveform, can adopt a configuration.
According to this configuration, the projection waveform in the scanning direction (or sub-scanning line) of a defect that appears as a “line” parallel to the scanning line (or sub-scanning line) is the scanning direction (or sub-operation) of the defect that appears as a “point”. (Direction) is larger than the projected waveform. Therefore, it is possible to quantitatively determine whether or not the type of defect of the test pattern is “point” or “line”.
In addition, when the defect type is “fog” and accompanied by chopping (hereinafter simply referred to as chopping fogging), the chopping has a strong tendency to repeatedly appear at a substantially constant short period, and therefore, a specific frequency region of the projected waveform. The spectral intensity at appears strongly. Therefore, it is possible to quantitatively determine whether or not the type of the defect included in the test pattern is “fogging” accompanied by fine cutting based on the maximum value of the spectral intensity in the specific frequency region of the projected waveform.
Further, when the defect type is a line, the “line” has a strong tendency to appear without periodicity, and thus the standard deviation of the spectrum intensity in the specific frequency region of the projected waveform becomes large. Therefore, it is possible to quantitatively determine whether the type of defect of the test pattern is “fogging” or “line” based on the standard deviation of the spectral intensity in the specific frequency region of the projected waveform.
Further, the vertical lines, which are linear defects parallel to the sub-scanning direction, tend to show a certain periodicity in the projection waveform in the sub-scanning direction and do not show the periodicity in the projection waveform in the main scanning direction. Therefore, by determining whether or not the standard deviation of the spectral intensity in the specific frequency region is the spectrum of the projected waveform in the main scanning direction, it can be determined whether the defect type is a vertical line or a horizontal line. I can judge.
Thus, for example, as compared with the case where the operator is to obtain qualitatively these feature values of the defect, it is possible not only to reduce the labor of an operator, identification accuracy of the defect type group guess the defect type set of operators Unaffected by superiority or inferior ability.

The fault diagnosis method according to the present invention specifies a defect type set having a defect type as an element, which is a type of defect included in the defect image information, from defect image information that is information acquired from an image having a defect output by itself. based in the first calculation step of calculating a current Gotoku symptom value, to the collector Gotoku symptoms value calculated by the central characteristic value and the first calculating step calculated in advance for each defect type set as the center of the defect type group Te, a first specifying step of specifying the corresponding defect type group mainly feature values used in those smallest among the calculated distance, from the defect image information, another species for specifying a defect type a second calculation step of calculating a feature value on the basis of the species-specific feature values calculated by the second calculating step, individual and specific conditions corresponding respectively to the defect type group identified by the first identifying step The disconnection process is characterized by comprising a second specifying step of specifying a defect type having defective image information.
According to this method, it is possible to identify to quantitatively and uniform defect type group by using a condensing Gotoku symptoms value. Thus, for example, the operator guesses a defect in the test pattern, and qualitatively or individually obtains and analyzes the value that characterizes the set for each defect type set having the estimated defect type as an element. Compared to the identification method, not only can the operator's labor be reduced, but the accuracy of identifying the defect type set is not affected by the superiority or inferiority of the operator's ability to infer the defect type set.
In addition, according to this configuration, in order to specify the defect type individually and specifically after the defect type set is uniformly specified by the cluster analysis, for example, a method of uniformly specifying the defect type only by the cluster analysis and Compared with this, it is possible to improve the accuracy of specifying the defect type.

According to the present invention, since the defect type is specified by the two-layer process of the clustering process and the individual specific condition determination process, the defect type is quantitatively and highly accurate based on the defect feature value obtained uniformly. Can be identified .

DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, exemplary embodiments of the invention will be described with reference to the accompanying drawings. FIG. 1 is a block diagram showing an embodiment of an image forming apparatus 100 of the present invention.
The image forming apparatus 100 according to the present invention includes an image reading unit 110, a print engine unit 120, a sensor unit 130, a failure diagnosis information input unit 140, a failure diagnosis unit 150, and a control unit 160.

  The image reading unit 110 is configured by an optical reading device such as a scanner, for example, and reads a document DC to acquire image information. The print engine unit 120 forms and outputs a read image or an image instructed to be printed. In particular, when the image reading unit 110 acquires image information from the document DC having an image defect output from the print engine unit 120, the image information is referred to as defect image information.

  The sensor unit 130 includes a sensor group. The sensor group obtains information related to the internal state of the apparatus such as paper passage time, drive current, internal temperature, and humidity.

  The failure diagnosis information input unit 140 includes, for example, a touch panel, a pointing device, or a keyboard, and receives information necessary for failure diagnosis. The failure diagnosis unit 150 performs failure diagnosis of the image forming apparatus 100 based on each information acquired by the failure diagnosis information input unit 140.

  The controller 160 includes, for example, an arithmetic device such as a CPU, a storage device such as a RAM, a recording device such as a ROM, and a program for controlling them. The control unit 160 controls the image reading unit 110, the print engine unit 120, the sensor unit 130, the failure diagnosis information input unit 140, and the failure diagnosis unit 150.

Next, the configuration of the failure diagnosis unit 150 will be described with reference to FIG. FIG. 2 is a block diagram illustrating a configuration example of the failure diagnosis unit 150.
The failure diagnosis unit 150 includes a defect type determination unit 1510, a feature amount extraction unit 1520 that is a feature amount extraction unit, an internal state information acquisition unit 1530 that is an internal state information acquisition unit, a failure probability inference unit 1540, and an additional operation information acquisition unit 1550. , And a diagnosis result notifying unit 1560 and the like.

  The defect type determination unit 1510 receives image information from the image acquisition unit 110. Next, the defect type determination unit 1510 determines the type of image defect based on the received image information. Thereafter, the determined defect type is transmitted to the control unit 160, the feature amount extraction unit 1520, and the failure probability inference unit 1540.

  The feature amount extraction unit 1520 extracts various feature amounts from the defect type determination result of the defect type determination unit 1510. Here, feature quantities such as shape, size, density, contour state, defect occurrence direction, and periodicity are extracted according to the type of defect.

  The internal state information acquisition unit 1530 includes a component state information acquisition unit 1531, a history information acquisition unit 1532, and an environment information acquisition unit 1533, and acquires various information inside the image forming apparatus 100 acquired from the sensor unit 130. To do. Thereafter, the acquired information is transmitted to the failure probability inference unit 1540.

  The component state information acquisition unit 1531 acquires component information indicating the operation state of each component based on the internal state information of the image forming apparatus 100 acquired from the sensor unit 130 as observation data information.

  The history information acquisition unit 1532 acquires the monitoring result of the usage status of the image forming apparatus 100 as history information. The environment information acquisition unit 1533 directly acquires environment information inside the image forming apparatus 100 or acquires environment information inside the image forming apparatus 100 acquired from the sensor unit 130.

  According to this configuration, the feature amount extraction and the acquisition of the internal state information are performed without the user, so that it is possible to improve the efficiency by eliminating the trouble of inputting the defect information one by one by the user, and there is no specialized knowledge about the device. However, detailed and accurate diagnosis is possible.

  The failure probability inference unit 1521 determines the defect type based on information obtained by the component state information acquisition unit 1531, the history information acquisition unit 1532, the environment information acquisition unit 1533, the feature amount extraction unit 1520, and the additional operation information acquisition unit 1550. Based on the diagnosis model corresponding to the defect type determined by the unit 1510, the failure probability of the failure cause on each model is calculated.

Further, the failure probability inference unit 1540 includes a failure candidate detection unit 1541, an inference engine 1542, a diagnostic model 1543, and the like.
The failure candidate detection unit 1541 narrows down failure cause candidates based on a failure cause probability calculated by an inference engine 1542 described later.

  The inference engine 1542 acquires from the internal operation information acquisition unit 1530 information acquired from the internal state information acquisition unit 1530, the probability that each cause candidate causing the failure will be the main cause of the failure that has occurred (failure cause probability). And the information acquired from the feature amount extraction unit 1520.

  The diagnosis model 1543 is one or a plurality of failure diagnosis models described later, and the diagnosis model corresponding to the defect type determined by the defect type determination unit 1510 is used to calculate the probability of the cause of the failure.

  Here, a Bayesian network is used as the inference engine 1542 for calculating the failure cause probability. This Bayesian network represents a problem area where the causal relationship is complex, so the causal relationship between multiple variables is sequentially connected and expressed as a network with a graph structure, and the dependency relationship between variables is represented by a directed graph. It is a thing. The fault diagnosis model in the present invention is constructed using this Bayesian network.

  Note that the fault diagnosis model, the image forming apparatus 100, and the like of the present invention can use conventional fault diagnosis models, image forming apparatuses, and the like.

The additional operation information acquisition unit 1550 acquires failure information in a state where the operation conditions differ depending on the user operation.
The diagnosis result notification unit 1560 is configured by a display device such as a control panel, for example, and notifies the user of the failure cause candidates extracted by the failure candidate extraction unit 1541.

  Next, the configuration of the defect type determination unit 1510 will be described with reference to FIG. FIG. 3 is a diagram for explaining a configuration example of the defect type determination unit 1510.

  The defect type determination unit 1510 is a set determination value calculation unit 1511 that is a set determination value calculation unit, a set determination unit 1512 that is a set determination unit, a type determination value calculation unit 1513 that is a type determination value calculation unit, and a type determination unit. It comprises a type determination unit 1514 and a concealment unit 1515 that is a concealment unit.

  The set determination value calculation unit 1511 is based on the defect image information which is multi-value image information acquired by the image acquisition unit 110 from the document DC having defects output by the print engine unit 120 itself, and the type of defect included in the defect image information. A set determination feature value for determining a defect type set having a defect type as an element is calculated.

  In addition, as will be described later, the set determination value calculation unit 1511 is a multi-value defect image in which a part of the defect types included in the image information is concealed by the concealment unit 1515 instead of the defect image information acquired by the image acquisition unit 110. A configuration for calculating the set determination feature value based on the information can be employed.

  Specifically, the set determination value calculation unit 1511 uses, as the set determination feature value, the integrated value of the gradation value histogram and the high gradation value that is the sum of the waveform widths where the gradation value exceeds the threshold in the gradation value projection waveform. Waveform width, standard deviation of gradation value histogram and standard deviation of gradation value projection waveform, number of peaks of gradation value projection waveform, maximum value of gradation value projection waveform, and intensity of frequency component of gradation value projection waveform Is calculated. The set determination value calculation unit 1511 transmits the calculated set determination feature value to the set determination unit 1512. As will be described later, the set determination value calculation unit 1511 transmits the calculated set determination feature value to the control unit 160 when necessary.

  The set determination unit 1512 receives the set determination feature value calculated by the set determination value calculation unit 1511. Next, the set determination unit 1512 determines a defect type set by clustering processing based on the received set determination feature value. Thereafter, the set determination unit 1512 transmits the determined defect type set to the type determination value calculation unit 1513, the concealment unit 1515, the feature amount extraction unit 1520, the failure probability inference unit 1540, and the control unit 160.

  The type determination value calculation unit 1513 receives the defect type set from the set determination unit 1512. In addition, defect image information which is multi-value image information acquired by the image acquisition unit 110 is received and binarized. Next, the type determination value calculation unit 1513 calculates a type determination feature value for determining a defect type, which is a type of defect included in the defect image information, from the defect image information. Thereafter, the calculated type determination feature value and defect type set are transmitted to the type determination value calculation unit 1513.

  Further, as will be described later, the type determination value calculation unit 1513 is a multi-value defect image in which a part of the defect types included in the image information is concealed by the concealment unit 1515 instead of the defect image information acquired by the image acquisition unit 110 A configuration can be adopted in which the information is binarized and the type determination feature value is calculated.

  Specifically, the type determination value calculation unit 1513 calculates the maximum value of the projection waveform, the maximum value of the spectrum intensity in the specific frequency region of the projection waveform, and the standard deviation of the spectrum intensity of the projection waveform as the type determination feature value. To do. The type determination value calculation unit 1513 transmits the calculated type determination feature value to the type determination unit 1514.

  Based on the type determination feature value received from the type determination value calculation unit 1513, the type determination unit 1514 performs defect condition information processing for each piece of defect image information through individual specific condition determination processing respectively corresponding to the defect type set determined by the set determination unit. Determining the type of defect possessed. Next, the type determination unit 1514 transmits the determined defect type to the concealment unit 1515, the feature amount extraction unit 1520, and the failure probability inference unit 1540.

  The concealing unit 1515 conceals defects belonging to the defect type determined by the set determining unit 1512 or the type determining unit 1514 from the defect image information. Thereafter, the concealed defect image information is transmitted to the set determination value calculation unit 1511 and the type determination value calculation unit 1513.

  The concealing unit 1515 conceals the defect image information acquired by the image acquiring unit 110 according to the defect type determined by the set determining unit 1512 or the type determining value calculating unit 1513, or the concealing unit 1515 itself conceals the defect. The defect included in the defect image information is concealed. Thereafter, the newly concealed defect image is transmitted to the set determination value calculation unit 1511 or the type determination value calculation unit 1513.

  Next, the configuration of a Bayesian network in the case of performing a fault diagnosis of an image defect system will be described with reference to FIG. FIG. 4 is a diagram conceptually illustrating a configuration example of a Bayesian network in the case of performing a fault diagnosis of an image defect system.

  As shown in the figure, this Bayesian network includes a failure cause node ND0 that represents a cause of causing an image defect, a component state node ND1 that represents state information of members (components) constituting the image forming apparatus, and the image forming apparatus 100. History information node ND2 representing the history information, environment information node ND3 representing the surrounding environment information where the image forming apparatus 100 is installed, observation state node ND4 representing the state information of the image quality defect, and the follow-up test result obtained by the user operation It includes a user operation node ND5 representing information and a defect type node ND6.

  The failure cause node ND0 is a node representing a cause causing an image defect, and the probability of this portion is calculated to determine whether or not there is a failure. Each node stores a probability table in which probability data representing the strength of the causal relationship is collected. The initial value of this probability data can be determined using data at the time of past failure occurrence or MTBF (Mean Time Between Failure) of the part.

  The component state node ND1 is a node representing a component state, and is information acquired from the sensor unit 130 that observes the component state. Such information includes information such as component temperature, applied voltage, patch density, and remaining amount of color material (for example, toner).

  The history information node ND2 represents the usage status of the image forming apparatus 100. For example, a history of the number of printed sheets for each component is used. This number of prints directly affects the state of the component, such as component wear and deterioration.

  The environment information node ND3 is an ambient environment condition that affects the state of the component, and in this embodiment, the temperature and humidity correspond to this. Temperature and humidity affect the image forming conditions and operating conditions of each component.

  The observation state node ND4 represents the observation state of defects generated in the output image, and is information observed and input by the user. For example, there is information such as the shape, size, density, contour, orientation, position, periodicity, and generation area of the defect.

  The user operation node ND5 is information that causes the image forming apparatus 100 to perform the same processing by changing the operation condition, and includes information about the changed operation condition.

  The defect type node ND6 represents the type of image defect and includes information such as lines, dots, white spots, and density unevenness. First, after determining the type of image defect that has occurred and confirming the state of this node, information on other nodes (ND1 to ND5) is appropriately input to make a diagnosis, and the cause of the failure is estimated.

  These nodes are connected so as to have a relationship of “cause” → “result”. For example, the relation between the “failure cause node” and the “observation state node ND4” is based on the “observation state (light concentration, lightness) indicated by the“ observation state node ND4 ”based on the“ cause ”indicated by the“ failure cause node ”. , Belt-like, etc.) "appears. On the other hand, the relationship between the “history information node ND2” and the “cause node” is “cause” (part deterioration, etc.) based on “history information based state (large number of copies, long operation years, etc.)”. This relationship holds.

  Next, a specific example of a failure diagnosis model in the failure diagnosis system will be described with reference to FIG. FIG. 5 is a diagram illustrating an example of a Bayesian network when a black line is generated in a configuration example of failure diagnosis due to an image defect.

  As shown in the figure, the nodes are connected so as to have a relationship of “cause” → “result”. For example, the relationship between “drum scratches” and “line width information” is such that “line width information” such as the occurrence of a thin line based on “drum scratches” appears.

  On the other hand, the relationship between “feed number history information” and “fuser” is based on the state based on “feed number” (the number of feeds is more than one), and the possibility of black line generation due to “fuser” deterioration increases. It holds.

  The initial value of the probability data of each node is determined based on, for example, past data. Thereafter, the probability of each node may be periodically updated based on market trouble statistical data such as the replacement frequency of parts and the frequency of occurrence of defects. Further, nodes representing image defect features such as “line width information”, “periodic information”, and “occurrence location information” in FIG. 5 are based on the feature amounts obtained by the feature amount extraction unit 1520 in FIG. The state is determined.

  Next, control processing executed by the control unit 160 will be described with reference to FIG. FIG. 6 is a flowchart illustrating an example of a control process executed by the control unit 160.

  First, the control unit 160 receives a command to output a test pattern when shifting to the failure diagnosis mode, and transmits a command to output a test pattern to the print engine unit 120 (step ST01).

  The test pattern output here is stored in advance in the print engine unit 120 shown in FIG. 1, and when the cause of the failure is a component of the print engine unit 120, the defect is reproduced in the test pattern. However, when there is a cause in a part of the image reading unit 110 such as a defect only at the time of copying, the defect is not reproduced in the test pattern. However, when there is a cause in the components of the image reading unit 110, when a test pattern is set in the image reading unit 110 and an output image is read, a defect appears in the read image. Accordingly, before reading the output image, an inquiry is made from the operation screen as to whether or not a defect occurs only at the time of copying, and the user can selectively input the information. Then, the selected information is acquired from the additional operation information acquisition unit 260 and input to the failure probability inference unit 207. When the test pattern is discharged from the print engine unit 120 of the image forming apparatus 100, the test pattern is set in the image reading unit 110, and the user inputs a command for instructing reading of the test pattern.

  Next, the control unit 160 receives the read command input by the user, and notifies the image acquisition unit 110 in FIG. 1 of the read command (step ST02).

  Next, the control unit 160 compares the read image with a reference image previously stored in the apparatus in the image defect detection unit 204 included in the failure diagnosis unit 150 to check whether there is an image defect (step ST03). . If it is determined that there is a defect, the control unit 160 performs the process in step ST05. If not, the controller 160 has been caused by an accident that has occurred accidentally, or has undergone some processing before the test pattern is output. Therefore, the user is notified of this through the operation screen and the process is terminated (step ST04-N).

  In step ST04, when it is determined that there is a defect (step ST04-Y), control unit 160 notifies the defect type determination unit 1510 of an execution command so as to determine the defect type (step ST05). Next, the control unit 160 selects a diagnostic model 1543 corresponding to the defect type determined by the defect type determination unit 1510 and notifies the failure probability inference unit 1540 (step ST06).

  Thereafter, control unit 160 notifies execution of feature amount extraction unit 1520 (step ST07). Next, the control unit 160 further includes status information of each component constituting the image forming apparatus 100, history information such as a counter value indicating the number of printed sheets for each component, and environmental information such as temperature and humidity inside the apparatus. In order to acquire various data necessary for failure diagnosis, an execution notification is sent to the component state information acquisition unit 1531, the history information acquisition unit 1532, and the environment information acquisition unit 1533 (step ST08).

  Thereafter, the control unit 160 notifies the inference engine 1542 of execution in order to calculate the occurrence probability of each failure cause based on the diagnosis model selected in step ST06 and the information acquired in steps ST07 and ST08 ( Step ST09).

  Next, based on the probability calculated in step ST09, the control unit 160 executes the failure candidate extraction unit 1541 to extract failure causes for the number of candidates specified from the higher probability of causing a failure. Notification is performed (step ST10). Note that the number of candidates may be set in advance, or an arbitrary number may be input and designated before candidate extraction.

  Thereafter, the control unit 160 controls the extracted failure cause candidate to be displayed on a display device such as a control panel by the diagnosis result notification unit 1560 and notified to the user (step ST11).

  Next, the control unit 160 determines whether or not failure cause candidates can be narrowed according to whether there is additional test result information (step ST12). When it is determined that the failure cause candidates can be narrowed down, control unit 160 ends the process, and otherwise, executes the process of step ST13.

  If it demonstrates in detail, in such an automatic determination process, a failure cause candidate may not necessarily be narrowed down to one at this time. Therefore, if the failure cause candidates cannot be narrowed down at this point, the user selects additional operation items necessary for further failure diagnosis from the operation screen.

  The control unit 160 issues an execution command to the print engine unit 120 so as to change the operation condition of the image forming apparatus 100 according to the selected item and re-output the image. Then, the user inputs information on the follow-up test result from the operation screen. The additional operation at this time is enlargement / reduction of an image, output of a test pattern held at each position of the image path, or the like, and examines whether or not a defect occurrence state has changed. Therefore, the follow-up test result is of a level that allows the user to easily input according to the question on the operation screen. Therefore, the control unit 160 determines that failure cause candidates cannot be narrowed down by receiving the added information.

  In step ST12, when determining that there is additional test result information, the control unit 160 transmits the received additional test result information to the inference engine 1542 (step ST13). Then, the added information and previously input information are combined to recalculate the failure cause probability, and the process returns to step ST09 to narrow down the failure candidates from the result, and the above processing is repeated.

  Next, an example of the process of determining the defect type shown in step ST05 of FIG. 6 (hereinafter simply referred to as the defect type determination process) will be described. FIG. 7 is a flowchart illustrating an example of defect type determination processing executed by the control unit 160.

  First, the control unit 160 notifies the set determination value calculation unit 1511 to execute the set determination feature value calculation process (step ST101). The set determination feature value calculation process will be described later.

  Next, the control unit 160 notifies the set determination unit 1512 to execute the primary stratification process (clustering process) (step ST102). The primary stratification process (clustering process) will be described later.

  After that, the control unit 160 determines whether the defect type set that is the determination result from the set determination unit 1512 is a set including only one element (step ST103). When the control unit 160 determines that the defect type set is a set including only one element, the control unit 160 determines that the element of the defect type set is the defect type included in the defect image information, and then ends the process. In that case, the process of step ST104 is executed.

  Next, in step ST103, when the control unit 160 determines that the defect type set is not a set of only one element, the control unit 160 notifies the type determination value calculation unit 1513 to execute the type determination feature value calculation process. (Step ST104). The type determination feature value calculation process will be described later.

  Thereafter, control unit 160 notifies execution to type determination unit 1514 so as to execute the secondary stratification process (individual comparison process) (step ST105). The secondary layer processing (individual comparison processing) will be described later. Thereafter, the control unit 160 ends the process.

According to this configuration, the defect type set can be determined quantitatively and uniformly using the set determination feature value. Thus, for example, the operator guesses a defect in the test pattern, and qualitatively or individually obtains and analyzes the value that characterizes the set for each defect type set having the estimated defect type as an element. Compared to the identifying method, not only the operator's labor can be reduced, but also the defect type set discrimination accuracy is not affected by the superiority or inferiority of the operator's ability to infer the defect type set.
In addition, according to this configuration, in order to determine the defect type individually and specifically after the defect type set is uniformly determined by the cluster analysis, for example, a method of uniformly determining the defect type only by the cluster analysis Compared to the defect type discrimination accuracy can be improved.

  Next, the set determination feature value calculation process shown in step ST101 of FIG. 7 will be described. FIG. 8 is a flowchart illustrating an example of a set determination feature value calculation process executed by the set determination value calculation unit 1511.

First, the set determination value calculation unit 1511 calculates the following set determination value from the multivalued defect image information acquired by the image acquisition unit 110.
The set determination value calculation unit 1511 calculates the integrated value of the gradation value histogram of the defect image information (hereinafter simply referred to as the gradation value histogram integrated value) (step ST201).

Next, the set determination value calculation unit 1511 calculates a high gradation value waveform width that is the number of pixels whose gradation value exceeds the threshold in the gradation value projection waveform (step ST202). The high gradation value waveform width will be described later.

  Thereafter, the set determination value calculation unit 1511 calculates the standard deviation of the gradation value histogram (hereinafter simply referred to as gradation value histogram standard deviation) (step ST203).

  Next, the set determination value calculation unit 1511 calculates the standard deviation of the gradation value projection waveform (hereinafter simply referred to as the gradation value projection waveform standard deviation) (step ST204).

  Thereafter, the set determination value calculation unit 1511 calculates the number of peaks of the gradation value projection waveform (hereinafter simply referred to as the gradation value projection waveform peak number) (step ST205).

  Next, the set determination value calculation unit 1511 calculates the maximum value of the gradation value projection waveform (hereinafter simply referred to as the gradation value projection waveform maximum value) (step ST206).

Thereafter, the set determination value calculation unit 1511 calculates the intensity of the frequency component of the gradation value projection waveform (hereinafter simply referred to as frequency component intensity) (step ST207).
Next, the set determination value calculation unit 1511 ends the process.

  Note that the set determination value calculation unit 1511 causes the print engine unit 120 to print the original DC having the defect image information and the defect that is the source of the defect image information, instead of the multi-value defect image information acquired by the image acquisition unit 110. It is possible to employ a configuration in which the feature value is calculated from image information of a multi-value difference image with the image information used for output.

  As will be described later, the set determination value calculation unit 1511 can employ an embodiment in which the image information concealed by the concealment unit 1515 is used instead of the defect image information acquired by the image acquisition unit 110.

  According to this configuration, the size of the defect area is increased by a certain width or more in the scanning direction (or sub-operation direction) due to the number of high gradation value pixels of the gradation value projection waveform due to the integral value of the gradation value histogram. The width of the sub-scanning direction (or the operation direction) varies depending on the standard deviation of the gradation value histogram and the standard deviation of the gradation value projection waveform. The number of defects of the defect set is such that the magnitude of the defect gradation value is based on the maximum value of the gradation value projection waveform, and the presence or absence of the periodic occurrence of the defect is based on the intensity of the frequency component of the gradation value projection waveform. Feature values that quantitatively represent features can be obtained uniformly. Therefore, “line / vertical line” that is a defect type set having “point” and “vertical line” as defect elements, and a defect type set that has “point” and “horizontal line” that are defect types as elements. Only “line / horizontal line”, defect type “cover”, “horizontal line” and “vertical line” as defect element set “vertical line / horizontal line / cover”, defect type “vertical line” Quantify which set of defects the test pattern belongs to, the “vertical line” that is the defect type set as the element, or the “horizontal line” that is the defect type set that has only the defect type “horizontal line” as the element And can be determined uniformly. Thus, for example, the operator guesses a defect in the test pattern, and qualitatively or individually obtains and analyzes the value that characterizes the set for each defect type set having the estimated defect type as an element. Compared to the identifying method, not only the operator's labor can be reduced, but also the defect type set discrimination accuracy is not affected by the superiority or inferiority of the operator's ability to infer the defect type set.

Here, the high gradation value waveform width calculated in step ST202 shown in FIG. 8 will be described with reference to FIG. FIG. 9 is a diagram for explaining the relationship between the high gradation value waveform width and the threshold value.
The upper diagram of FIG. 9 is defect image information DI acquired by the image acquisition unit 110 as image information on the document DC having defects output from the print engine unit 120.

  The defect image information DI is defect image information obtained by the image acquisition unit 110 acquiring the document DC having a defect output from the print engine unit 120 as image information. The defect image information DI has an image defect composed of two lines. Further, these two lines have the highest density at the center, and the density decreases as the distance from the center increases. The defect image information DI is set with coordinate axes in the main scanning direction DM and the sub-scanning direction DS.

  The lower diagram of FIG. 9 is a diagram showing the projection waveform in the main scanning direction DM of the defect image information DI shown in the upper diagram. The vertical axis represents the average of the gradation values in the sub-scanning direction for each unit distance (for example, pixel) in the scanning direction. The horizontal axis is the coordinate axis in the main scanning direction DM of the defect image information DI shown in the above figure. The lower diagram in FIG. 9 shows that the average gradation value at the center of the two lines is high and decreases as the distance from the center is off.

In the lower diagram of FIG. 9, a predetermined threshold value is set in advance, and the high gradation value waveform width is d1 and d2 which are the widths of the portion where the average gradation value exceeds the threshold value in the gradation value projection waveform. Is calculated as the sum of

  Next, the primary stratification process (clustering process) shown in step ST102 of FIG. 7 will be described with reference to FIG. FIG. 10 is a diagram for explaining the relationship between clusters and defect type sets.

  FIG. 10 shows a vertical line / horizontal line / fogging cluster CVHO composed of a coordinate group represented by a set determination feature value calculated when the type of defect observed in the past is a vertical line, horizontal line, or fogging. , A horizontal line cluster CH composed of a coordinate group represented by a set determination feature value calculated when the defect type is only a horizontal line, and a set determination feature value calculated when the defect type is only a vertical line A vertical line cluster CV composed of coordinate groups represented by the following: vertical line / point cluster CVD composed of coordinate groups represented by set determination feature values calculated when the type of defect is a vertical line or point , And a horizontal line / point cluster CHD composed of a coordinate group represented by a set determination feature value calculated when the type of defect is a horizontal line or a point.

  Also, the vertical line / horizontal line / cover cluster CVHO is the vertical line / horizontal line / cover set GHVO, the horizontal line cluster CH is the horizontal line set GH, the vertical line cluster CV is the vertical line set GV, and the vertical line / color point cluster is the vertical line. / Point set GVD and horizontal line / point cluster respectively correspond to horizontal line / point set GHD.

  Therefore, the vertical line / horizontal line / fogging set GVHO as the defect type set includes the vertical line, horizontal line, or fogging as the defect type as elements, and the horizontal line set GH includes only the horizontal line as the defect type as a vertical line set GV. Is a defect type vertical line / point set GVD is a defect type vertical line and color point, and a defect type horizontal line / point set GHD is a defect type. The horizontal lines and color points that are the types are used as elements.

Although FIG. 10 is displayed with a three-dimensional coordinate space for convenience of explanation, it is not actually limited to a three-dimensional space. In this embodiment, the tone value histogram integrated value, the high tone value waveform width, the tone value histogram standard deviation, the tone value projection waveform standard deviation, the tone value projection waveform peak number, the tone, which are adopted as the set determination feature values. A multidimensional coordinate space represented by each coordinate axis corresponding to the maximum value projection waveform maximum value and the frequency component intensity is used.

Next, referring to the primary layer by the process shown in step ST102 in FIG. 7 to FIG. 11 for (clustering processing) will be described. FIG. 11 is a diagram for explaining the relationship between the set determination feature value vector composed of the set determination feature value calculated by the set determination value calculation unit 1511 and the cluster shown in FIG.

The feature value table TB displays a set determination feature value corresponding to each element of the set determination feature value vector X. In order, the elements x1, x2, x3, x4, x5, x6, x7 are the tone value histogram integral value, the high tone value waveform width, and the tone value histogram which are the set judgment feature values calculated by the set judgment value calculation unit 1511. It corresponds to standard deviation, gradation value projection waveform standard deviation, gradation value projection waveform peak number, gradation value projection waveform maximum value, and frequency component intensity.

The defect type set table TS displays the relationship between the vertical line / point, horizontal line / point, vertical line, horizontal line, vertical line / horizontal line / fogging cluster, and the set determination feature value vector X.
A vertical line / point cluster corresponds to a vertical line / point set. A vertical line / point set is not composed of a single defect. That is, since the vertical line / point set is composed of a plurality of elements, the defect type is not determined when it is determined that the defect belongs to the vertical line / point set.

  Next, the center vector C1 of the vertical line / point cluster includes a coordinate group represented by a set determination feature value of defects whose types observed in the past constituting the vertical line / point cluster are vertical lines or points. Calculated as the center of the smallest sphere.

  Next, the distance D1 between the vertical line / point cluster and the set determination feature value vector X is defined by the Mahalanobis distance between the center vector C1 and the set determination feature value vector X. The relationship between the horizontal line / point, vertical line, horizontal line, vertical line / horizontal line / fogging cluster and the set determination feature value vector is the same as the relationship between the color point / vertical line cluster and the set determination feature value vector, and will not be described.

  Next, the primary stratification process (clustering process) shown in step ST102 of FIG. 7 will be described with reference to FIG. FIG. 12 is a flowchart for explaining primary stratification processing executed by the set determination unit 1512.

  First, the set determination unit 1512 initializes a distance variable Di between all cluster center vectors Ci and the set determination feature value vector X (step ST301). However, i is an integer from 1 to n. N is the number of clusters.

  Next, the set determination unit 1512 determines whether there is a feature value center vector Ci (hereinafter simply referred to as an unprocessed feature value center vector Ci) that is not a processing target among the feature value center vectors Ci of all clusters. (Step ST302). The set determination unit 1512 performs the process of step ST303 when determining that the unprocessed feature value center vector Ci exists, and executes the process of step ST304 otherwise.

  In step ST302, when the set determination unit 1512 determines that there is an unprocessed feature value center vector Ci, the set determination unit 1512 calculates the distance between one of the unprocessed feature value center vectors Ci and the set determination feature value vector X. (Step ST303). Then, it returns to step ST302 and repeats the said process.

  In step ST302, when the set determination unit 1512 determines that the unprocessed feature value center vector Ci does not exist, the set determination unit 1512 sets the minimum one of the distances Di calculated in step ST303 as Dmin (step ST304). Next, the set determination unit 1512 acquires a defect type set corresponding to the feature value center vector Ci used to calculate the distance Dmin, and determines that the defect type is included in the defect image information (Step S1512). ST305). Thereafter, the set determination unit 1512 ends the process.

Next, a description will be given type determination characteristic value calculation process shown in step ST10 4 of FIG. 7 with reference to FIG. 13. FIG. 13 is a flowchart for explaining type determination feature value calculation processing executed by the type determination value calculation unit 1513.

  First, the type determination value calculation unit 1513 executes binarization processing (step ST401). That is, the type determination value calculation unit 1513 executes a process of binarizing the defect image information acquired by the image acquisition unit 110, or a document DC having a defect that is the basis of the defect image information and the defect image information. Is binarized with respect to the difference image from the image information used when the print engine unit 120 outputs.

  As will be described later, the type determination value calculation unit 1513 can employ an embodiment in which the image information concealed by the concealment unit 1515 is used instead of the defect image information acquired by the image acquisition unit 110.

  Next, the type determination value calculation unit 1513 executes a process of calculating the number of pixels from the image information binarized in step ST401 (hereinafter simply referred to as a pixel number calculation process) (step ST402).

  Thereafter, the type determination value calculation unit 1513 executes a process of calculating a projection waveform based on the number of pixels from the image information binarized in step ST401 (hereinafter simply referred to as a projection waveform calculation process) (step ST403).

  Next, the type determination value calculation unit 1513 executes a process of calculating the type determination feature value from the projection waveform based on the number of pixels (step ST404). That is, the type determination value calculation unit 1513 that has received the command calculates the maximum value of the projection waveform, the maximum value of the spectrum intensity in the specific frequency region of the projection waveform, and the standard deviation of the spectrum intensity of the projection waveform. Thereafter, the type determination value calculation unit 1513 ends the process.

  According to this configuration, the feature value calculated from the test pattern expressed in multiple values strongly expresses the feature related to the print density of the defect type as compared to the feature value calculated from one expressed in binary. It does not show the characteristics of the defect type shape very strongly. In addition, the feature value calculated from the test pattern expressed in binary does not express the feature related to the print density of the defect type too strongly compared to the feature value calculated from expressed in multiple values. It strongly expresses the characteristics related to the type of shape. Thereby, since the defect type can be determined from the print density and shape of the defect type, respectively, for example, the defect type determination accuracy is improved as compared with the case of determining the defect type from only the multi-value or only the binary value. be able to.

According to this configuration, the projection waveform in the scanning direction (or sub-scanning line) of a defect that appears as a “line” parallel to the scanning line (or sub-scanning line) is the scanning direction (or sub-operation) of the defect that appears as a “point”. (Direction) is larger than the projected waveform. Therefore, it is possible to quantitatively determine whether the type of the defect that the test pattern has is “point” or “line”.
In addition, when the defect type is “fog” and accompanied by chopping (hereinafter simply referred to as chopping fogging), the chopping has a strong tendency to repeatedly appear at a substantially constant short period, and therefore, a specific frequency region of the projected waveform. The spectral intensity at appears strongly. Therefore, it is possible to quantitatively determine whether or not the type of the defect included in the test pattern is “fogging” accompanied by fine cutting based on the maximum value of the spectral intensity in the specific frequency region of the projected waveform.
Further, when the defect type is a line, the “line” has a strong tendency to appear without periodicity, and thus the standard deviation of the spectrum intensity in the specific frequency region of the projected waveform becomes large. Therefore, it is possible to quantitatively determine whether the type of defect of the test pattern is “fogging” or “line” based on the standard deviation of the spectral intensity in the specific frequency region of the projected waveform.
Further, the vertical lines, which are linear defects parallel to the sub-scanning direction, tend to show a certain periodicity in the projection waveform in the sub-scanning direction and do not show the periodicity in the projection waveform in the main scanning direction. Therefore, it is possible to determine whether the defect type is a vertical line or a horizontal line by determining whether the standard deviation of the spectral intensity in the specific frequency region is the spectrum of the projected waveform in the main scanning direction. .
Therefore, for example, compared to the case where the operator qualitatively acquires these feature values of the defect, not only the operator's labor can be reduced, but also the defect type set discrimination accuracy can be estimated by the operator. Unaffected by superiority or inferior ability.

  Here, the relationship between the defect type and the spectrum intensity will be described with reference to FIGS. FIG. 14 is a diagram for explaining an example of a spectrum of defect image information whose defect type is a horizontal line.

  FIG. 14 shows an example of image defect information whose defect type is a horizontal line in order from the top, an example of the relationship between the spectral intensity of the main scanning projection waveform and the spectral intensity of the sub scanning projection waveform of the image defect information. ing.

  As shown in the middle diagram of FIG. 14, in the region between the specific frequencies fLys and fLye of the projection waveform in the sub-scanning direction of the image defect whose defect type is a horizontal line, the influence of the horizontal line as the image defect is a spectrum power curve. Tends to appear strongly as a discontinuous curve.

  Therefore, the standard deviation SLy_σ of the spectrum intensity between the specific frequency region fLys and fLye when the defect type is a horizontal line is a gentle spectrum power curve between the specific frequency region fLys and fLye, as will be described later. Compared with the case where the defect type that tends to appear as a continuous curve is a vertical line or a fog, it tends to show a larger value.

  Further, in the frequency region fHys to fHye having a value larger than the region between the specific frequencies fLys to fLye, the influence of the horizontal line that is an image defect does not appear, and the spectrum power curve is a gentle continuous curve, and the frequency It tends to decrease gradually with the increase of.

Next, the relationship between image defect information whose defect type is a horizontal line and the spectrum intensity of the main scanning projection waveform will be described with reference to the lower diagram of FIG.
As shown in the lower diagram of FIG. 14, in the region between the specific frequencies fLxs to fLxe of the projection waveform in the main scanning direction of the image defect information whose defect type is a horizontal line, the influence of the horizontal line that is an image defect does not appear. The power curve is a gentle continuous curve and tends to decrease as the frequency increases.

  Therefore, the standard deviation SLx_σ of the spectrum intensity between the specific frequency region fLxs and fLxe when the defect type is a horizontal line tends to be smaller than the spectrum power curve between the specific frequency region fLys and fLye. It is in.

  In addition, in the frequency region fHxs to fHxe having a value larger than the region between the specific frequencies fLxs to fLxe, the influence of the horizontal line which is an image defect does not appear, and the spectrum power curve is a frequency curve. It tends to decrease gradually with the increase.

Next, with reference to FIG. 15, the relationship between image defect information whose defect type is fogging and the spectral intensity of the projected waveform will be described.
FIG. 15 is an example of image defect information whose defect type is a thin fogging in order from the top, an example of the relationship between the spectral intensity of the main scanning projection waveform and the spectral intensity of the sub-scanning projection waveform of the image defect information. Represents.

  As shown in the middle diagram of FIG. 15, in the region between the specific frequencies fLys and fLye of the projection waveform in the sub-scanning direction of the image defect information whose defect type is a finely divided fog, the influence of the finely divided fog which is an image defect is spectrum · It does not appear in the power curve, but tends to appear as a gentle continuous curve.

  Therefore, the standard deviation SLy_σ of the spectral intensity between the specific frequency region fLys and fLye when the defect type is a thin fog is a tendency that the spectrum power curve between the specific frequency region fLys and fLye appears as a discontinuous curve. There is a tendency to show a smaller value than when a certain defect type is a horizontal line.

  In addition, in the frequency region fHys to fHye having a value larger than the region between the specific frequencies fLys to fLye, the influence of the fine fogging that is an image defect appears strongly, and the spectrum power curve is a mountain-shaped curve with a peak. Tend to show. This is because shreds tend to appear with a certain period.

  Therefore, the maximum value SHy_max of the spectral intensity between the specific frequency region fHys and fHye when the defect type is finely divided fog is continuous with a gentle spectrum power curve between the specific frequency region fHys and fHye, Compared to the case where the defect type that tends to appear as a curve that gradually decreases with increasing frequency is a horizontal line or solid fog, it tends to show a larger value.

Next, with reference to the lower part of FIG. 15, the relationship between the image defect whose defect type is a fogging and the spectral intensity of the main scanning projection waveform will be described.
As shown in the lower diagram of FIG. 15, in the region between the specific frequencies fLxs to fLxe of the projection waveform in the main scanning direction of the image defect whose defect type is a finely divided fog, the influence of the finely divided fog which is an image defect does not appear and the spectrum • The power curve is a gentle continuous curve that tends to decrease with increasing frequency.

  Therefore, the standard deviation SLx_σ of the spectral intensity between the specific frequency region fLxs and fLxe when the defect type is a thin fog is the same as the standard deviation SLy_σ of the spectral intensity between the specific frequency region fLys and fLye. Tend to show.

  In addition, the frequency region fHxs to fHxe having a value larger than the region between the specific frequencies fLxs to fLxe does not show the influence of fine fogging, which is an image defect, is a gentle continuous curve, and the spectrum curve increases the frequency. It tends to decrease gradually.

Next, with reference to FIG. 16, the relationship between the image defect whose defect type is solid fog and the spectral intensity of the projection waveform will be described.
FIG. 16 shows an example of image defect information whose defect type is a solid fog in order from the top, an example of the relationship between the spectral intensity of the main scanning projection waveform and the spectral intensity of the sub-scanning projection waveform of the image defect information. Represents.

  As shown in the middle diagram of FIG. 16, in the region between the specific frequencies fLys and fLye of the projection waveform in the sub-scanning direction of the image defect whose defect type is a solid fog, the influence of the solid fog as an image defect is spectral power.・ It does not appear on the curve, but tends to appear as a gentle continuous curve.

  Therefore, the standard deviation SLy_σ of the spectral intensity between the specific frequency region fLys and fLye when the defect type is solid fog tends to appear as a discontinuous curve of the spectral power curve between the specific frequency region fLys and fLye. There is a tendency to show a smaller value than when a certain defect type is a horizontal line.

  Also, the frequency region fHys to fHye having a value larger than the region between the specific frequencies fLys to fLye does not show the influence of solid fog as an image defect, and the spectrum power curve is a gentle continuous curve, It tends to decrease with increasing frequency.

  Therefore, the spectrum in the specific frequency region fHys to fHye when the defect type is solid fog tends to represent a mountain-like curve with a peak in the spectrum power curve between the specific frequency region fHys and fHye. There is a tendency to show a smaller value than in the case of a finely ridden fog.

Next, the relationship between the image defect whose defect type is a solid fog and the spectral intensity of the main scanning projection waveform will be described with reference to the lower diagram of FIG.
As shown in the lower diagram of FIG. 16, in the region between the specific frequencies fLxs to fLxe of the projected waveform in the main scanning direction of the image defect whose defect type is a solid fog, the influence of the solid fog as an image defect does not appear, and the spectrum • The power curve is a gentle continuous curve that tends to decrease with increasing frequency.

  Therefore, the standard deviation SLx_σ of the spectral intensity between the specific frequency region fLxs and fLxe when the defect type is a solid fog is set to a small value, similar to the standard deviation SLy_σ of the spectral intensity between the specific frequency region fLys and fLye. Tend to show.

  Further, the frequency region fHxs to fHxe having a value larger than the region between the specific frequencies fLxs to fLxe does not show the influence of solid fog as an image defect, and is a gentle continuous curve, and the spectrum power curve has a frequency. It tends to decrease gradually with the increase of.

  Therefore, the secondary layer processing (individual comparison processing) executed by the control unit 160 shows a characteristic value due to such a difference in defect type when the defect type set is a fog / vertical line / horizontal line set. The maximum value SHy_max of the spectral intensity in the specific frequency region of the projection waveform and the standard deviation SLx_σ or SLy_σ of the spectral intensity of the projection waveform, which are the type determination feature values, respectively, are set to a predetermined threshold SpecTh and SigmaMin or SigmaMax The defect type is determined as a fog, vertical line, or horizontal line by individually comparing whether or not it exceeds.

  Next, a relationship between a defect image whose defect type is a point or a vertical line and a projection waveform will be described with reference to FIG. FIG. 17 is a diagram for explaining a relationship between a defect image whose defect type is a point or a vertical line and a projection waveform.

  As shown in FIG. 17, the projection waveform in the main scanning direction obtained from the defect image whose defect type is a point tends to be smaller than the projection waveform in the main scanning direction obtained from the defect image whose defect type is a vertical line. is there. This is because a linear defect type parallel to the sub-scanning line is defined as a vertical line. Therefore, the projection waveform in the sub-scanning direction obtained from the defect image whose defect type is a point tends to be smaller than the projection waveform in the sub-scanning direction obtained from the defect image whose defect type is a vertical line.

  Therefore, the secondary stratification process (individual comparison process) executed by the control unit 160 is characterized by such a difference in defect type when the defect type set is a point / vertical line or a point / vertical line set. The defect type is a point, vertical line, or horizontal line by individually comparing whether or not the maximum value of the projected waveform, which is a type determination feature value indicating a specific value, exceeds a predetermined threshold BinMax Determine either of these.

Next, referring to secondary layer specific processing shown in step ST105 in FIG. 7 to FIG. 18 (individual comparison processing) will be described. FIG. 18 is a flowchart for explaining secondary stratification processing executed by the type determination unit 1514.

  First, the type determination unit 1514 determines whether the defect type set determined by the primary stratification process (clustering process) ordered to be executed by the set determination unit 1512 in step ST102 of FIG. 7 is a vertical line / horizontal line / fogging set. Judgment is made (step ST501). The type determination unit 1514 executes the process of step ST502 when determining that the defect type set is a vertical line / horizontal line / cover set, and otherwise executes the process of step ST506.

  In step ST501, when the type determination unit 1514 determines that the defect type set is a vertical line / horizontal line / fogging set, the projection waveform in the sub-scanning direction, which is the feature value extracted in step ST404 of FIG. It is determined whether the maximum value SHy_max of the spectrum intensity between the frequency domain fLys and fLye exceeds a predetermined threshold value SpecTh (step ST502). If it is determined that SHy_max exceeds SpecTh, type determination unit 1514 executes the process of step ST505, and otherwise executes the process of step ST503.

  In step ST502, if the type determination unit 1514 determines that the maximum value SHy_max does not exceed the threshold value SpectTh, the type determination unit 1514 standard deviation of the spectral intensity in the main scanning direction between the specific frequency region fLxs and fLxe. It is determined whether SLx_σ is smaller than a predetermined threshold SigmaMin and the standard deviation SLy_σ of the spectral intensity in the sub-scanning direction between the specific frequency region fLys and fLye is larger than the predetermined threshold SigmaMax (step ST503). .

  If it is determined that SLx_σ is smaller than SigmaMin and SLy_σ is larger than SigmaMax, type determination unit 1514 performs the process of step ST504, and otherwise performs the process of step ST511.

  In step ST503, when determining that SLx_σ is smaller than SigmaMin and SLy_σ is larger than SigmaMax, type determination unit 1514 determines whether SLx_σ is larger than SigmaMax and SLy_σ is smaller than SigmaMin. (Step ST504).

  When it is determined that SLx_σ is larger than SigmaMax and SLy_σ is smaller than SigmaMin, the process of step ST505 is executed, and otherwise, the process of step ST512 is executed.

  In step ST502, the type determination unit 1514 determines that the maximum value SHy_max exceeds the threshold value SpecTh, or in step ST504, the type determination unit 1514 has SLx_σ larger than SigmaMax and SLy_σ smaller than SigmaMin. If it is determined, the defect type is determined to be fogging (step ST505). Thereafter, the type determination unit 1514 ends the process.

  In step ST501, when the type determination unit 1514 determines that the defect type set is not a vertical line / horizontal line / cover set, the type determination unit 1514 determines whether the defect type set is a vertical line / point ( Step ST506). The type determination unit 1514 executes the process of step ST507 when determining that the defect type set is a vertical line / point, and otherwise executes the process of step ST509.

  In step ST506, when determining that the defect type set is a point / vertical line, the type determining unit 1514 determines whether the maximum value of the main scanning projection waveform is smaller than a predetermined threshold Binmax (step ST507). If it is determined that the maximum value of the main scanning projection waveform is smaller than the predetermined threshold value Binmax, type determination unit 1514 executes the process of step ST508, and otherwise executes the process of step ST513.

  In step ST507, if the type determination unit 1514 determines that the maximum value of the main scanning projection waveform is smaller than the predetermined threshold Binmax, the type determination unit 1514 determines the defect type to be a point (step ST508). Thereafter, the type determination unit 1514 ends the process.

  In step ST506, if the type determination unit 1514 determines that the defect type set is not a vertical line / point, it determines whether the maximum value of the sub-scanning projection waveform is smaller than a predetermined threshold Binmax (step ST509). If it is determined that the maximum value of the sub-scanning projection waveform is smaller than the predetermined threshold value Binmax, type determination unit 1514 executes the process of step ST510, and otherwise executes the process of step ST514.

  If the type determination unit 1514 determines in step ST509 that the maximum value of the sub-scanning projection waveform is smaller than the predetermined threshold Binmax, the type determination unit 1514 determines that the defect type is a point (step ST510). Thereafter, the type determination unit 1514 ends the process.

  In step ST503, the type determining unit 1514 determines that the defect type is a horizontal line when SLx_σ is greater than or equal to SigmaMin or SLy_σ is less than or equal to SigmaMax (step ST511). Thereafter, the type determination unit 1514 ends the process.

  In step ST504, the type determination unit 1514 determines that the defect type is a vertical line when SLx_σ is equal to or lower than SigmaMax or SLy_σ is equal to or higher than SigmaMin (step ST512). Thereafter, the type determination unit 1514 ends the process.

  In step ST507, the type determining unit 1514 determines that the defect type is a vertical line when determining that the maximum value of the main scanning projection waveform is equal to or greater than the predetermined threshold Binmax (step ST513). Thereafter, the type determination unit 1514 ends the process.

  In step ST509, the type determining unit 1514 determines that the defect type is a horizontal line when determining that the maximum value of the sub-scanning projection waveform is equal to or greater than a predetermined threshold Binmax (step ST514). Thereafter, the type determination unit 1514 ends the process.

  Next, another example of the defect type determination process shown in step ST05 of FIG. 6 will be described. FIG. 17 is a flowchart illustrating another example of defect type determination processing executed by control unit 160. This defect type determination process is different from the defect type determination process shown in FIG. 7 in that a plurality of defect types can be determined.

  First, the control unit 160 executes the processes of steps ST601 to ST605. Steps ST601 to ST605 are the same as steps ST101 to ST105 shown in FIG.

  However, unlike step ST102 shown in FIG. 7, step ST602 is executed after step ST601 is executed or when it is determined in step ST610 that the distance between the feature value vector and the origin is smaller than the threshold. It is different.

  In addition, when it is determined in step ST603 that the defect type set is a set including only one element, or after executing step ST605, the control unit 160 executes the process of step ST606 without terminating the process. But it is different.

  In step ST603, when the control unit 160 determines that the defect type set is a set of only one element, or after executing step ST605, the control unit 160 specifies a defect position to the concealment unit 1515 (hereinafter simply referred to as “defect position set”). An execution instruction is issued to execute a defect position specifying process (step ST606). The defect position specifying process will be described later.

Next, control section 160 gives an execution instruction to conceal the defect position to concealing section 1515 (step ST607).
Thereafter, similarly to step ST601, control unit 160 notifies execution to set determination value calculation unit 1511 so as to execute the set determination feature value calculation process (step ST608). Note that the set determination value calculation unit 1511 executes the set determination feature value calculation process based on the image information concealed by the concealment unit 1515 and recalculates the set determination feature value, and follows the instruction in step ST601. This is different from the execution of the set determination feature value calculation process.

  Thereafter, control section 160 calculates the distance between the set determination feature value vector composed of the set determination feature values calculated by set determination value calculation section 1511 and the origin (step ST609).

  Next, control section 160 determines whether or not the distance between the set determination feature value vector and the origin is smaller than a predetermined threshold (step ST610). If it is determined that the distance between the feature value vector and the origin is smaller than the predetermined threshold, the control unit 160 ends the process. If not, the control unit 160 returns to step ST602 and repeats the above process. In step ST602, the defect type set Are re-determined, the type determination feature value is recalculated in step ST604, and the defect type is re-determined in step ST605. The distance between the feature value vector and the origin is calculated using the Mahalanobis distance.

  According to this configuration, when the test pattern has a plurality of defects, and the plurality of defects are composed of a plurality of types of defects, the defect types that have already been determined to be the defect types constituting the defects are concealed. In order to perform re-determination, it is possible to determine a plurality of types of defects constituting the defect.

  According to this configuration, since the length of the set determination feature value is close to 0 when there is no defect, the length of the set determination feature value vector determines whether or not the test pattern has a defect. And can be judged in a consistent and uniform manner. Therefore, for example, it is possible not only to reduce the operator's labor as compared with the case where the operator determines the presence / absence of a defect qualitatively or individually, but also the defect type set discrimination accuracy estimates the operator's defect type set. Unaffected by superiority or inferior ability.

  Next, the defect position specifying process shown in step ST606 of FIG. 19 will be described with reference to FIG. FIG. 20 is a flowchart for explaining an example of the defect position specifying process executed by the concealing unit.

  First, concealment unit 1515 determines whether the defect type determined by set determination unit 1512 or type determination unit 1514 is a horizontal line (step ST701). The concealing unit 1515 executes the process of step ST702 when determining that the defect type is a horizontal line, and executes the process of step ST704 otherwise.

  In step ST701, when the concealing unit 1515 determines that the defect type is a horizontal line, the main scanning direction projection waveform is larger than the predetermined threshold value 11 and the sub-scanning direction projection waveform is larger than the predetermined threshold value 12. A large area is extracted (step ST702). The threshold value 11 and the threshold value 12 are predetermined threshold values in the main scanning direction and the sub-scanning direction, respectively, when the defect type is a horizontal line.

  Thereafter, concealment unit 1515 executes mask range determination processing 1 (step ST703). Specifically, a range obtained by expanding the region extracted in step ST702 by a predetermined width in the sub-scanning direction is determined as a mask range. The reason for expanding in the sub-scanning direction is that the defect type is determined to be a horizontal line. Thereafter, the concealment unit 1515 ends the process.

  In step ST701, when the concealing unit 1515 determines that the defect type is not a horizontal line, it determines whether the defect type is a vertical line (step ST704). The concealing unit 1515 executes the process of step ST705 when determining that the defect type is a vertical line, and executes the process of step ST707 otherwise.

  In step ST704, when the concealment unit 1515 determines that the defect type is a vertical line, the main scanning direction projection waveform is larger than the predetermined threshold value 21 and the sub-scanning direction projection waveform is higher than the predetermined threshold value 22. Is extracted (step ST705). The threshold value 21 and the threshold value 22 are predetermined threshold values in the main scanning direction and the sub-scanning direction, respectively, when the defect type is a vertical line.

  Thereafter, concealment unit 1515 performs mask range determination processing 2 (step ST706). Specifically, a range obtained by expanding the area extracted in step ST705 by a predetermined width in the main scanning direction is determined as a mask range. The reason for expanding in the main scanning direction is that the defect type is determined to be a vertical line. Thereafter, the concealment unit 1515 ends the process.

  If concealment unit 1515 determines in step ST704 that the defect type is not a vertical line, it determines whether the defect type is a point (step ST707). When concealing unit 1515 determines that the defect type is a point, it executes the process of step ST708.

  In step ST707, when the concealment unit 1515 determines that the defect type is a point, the main scanning direction projection waveform is larger than the predetermined threshold 31 and the sub-scanning direction projection waveform is larger than the predetermined threshold 32. A large area is extracted (step ST708). The threshold value 31 and the threshold value 32 are predetermined threshold values in the main scanning direction and the sub-scanning direction, respectively, when the defect type is a point.

  Thereafter, concealment unit 1515 executes mask range determination processing 3 (step ST709). Specifically, a range obtained by expanding the region extracted in step ST705 by a predetermined width in the main scanning direction and the sub-scanning direction is determined as a mask range. The reason for expanding in the main scanning direction and the sub-scanning direction is that the defect type is determined to be a point. Thereafter, the concealment unit 1515 ends the process.

  In step ST707, when the concealment unit 1515 determines that the defect type is not a point, the description is omitted, but the determination process similar to that shown in steps ST701 and ST704 and the steps ST702 and ST705 are performed. A process for extracting the same region as shown is executed based on the threshold value corresponding to each defect type, and a mask range determination process similar to that shown in steps ST703 and ST706 is executed based on each defect type. Thereafter, the concealment unit 1515 ends the execution of the process.

  Next, with reference to FIGS. 21 and 22, a case will be described in which the defect type determination process shown in FIG. 19 is performed on defect image information having two defect types of points and horizontal lines. 21 and 22 are diagrams for explaining defect type determination processing for defect image information having two defect types, that is, a point and a horizontal line.

The upper diagram of FIG. 21 represents defect image information having two defect types, a point and a horizontal line.
The middle diagram of FIG. 21 shows the main scanning direction projection waveform and the sub-scanning direction projection waveform drawn by the concealing unit 1515 from the defect image information having two defect types of points and horizontal lines, and the set determination value calculation unit 1511 calculates. This represents a set determination feature value vector having the set determination feature value as an element.

  The middle diagram of FIG. 21 represents that the length of the set determination feature value vector X1 exceeds a predetermined threshold. Further, the case where the defect type is determined to be a horizontal line by the primary stratification process shown in step ST102 of FIG. 7 is shown.

  The concealing unit 1515 determines that the defect type is a horizontal line in step ST701 of the flowchart shown in FIG. get. As a result, the region included in the horizontal line defect can be acquired.

  The lower part of FIG. 21 shows a region including the region acquired in step ST702. This shows the mask range determination processing 1 in step ST703 executed by the concealment unit 1515. That is, the mask range determined by the concealing unit 1515 indicates that the region extracted in step ST702 shown in the middle of FIG. 19 is a range that is expanded by a predetermined width in the sub-scanning direction.

  The upper diagram of FIG. 22 shows a defect image after masking the range determined by the mask range determination process 1 by the concealing unit 1515. As a result of the horizontal line defect being concealed from the defect image information, only a point defect exists in the defect image information.

  The upper diagram of FIG. 22 shows the main scanning direction projection waveform and the sub-scanning direction projection waveform drawn by the concealing unit 1515 from the defect image information having the defect type of point, and the set determination value, as in the middle diagram of FIG. A set determination feature value vector X2 having the set determination feature value calculated by the calculation unit 1511 as an element is shown.

  The upper diagram of FIG. 22 shows that the length of the set determination feature value vector X2 exceeds a predetermined threshold. Further, the case where the defect type is determined to be a point by the primary stratification process shown in step ST102 of FIG. 7 is shown.

  The concealing unit 1515 determines that the defect type is a point in step ST701 in the flowchart shown in FIG. 20, and in step ST702, the concealment unit 1515 includes an area where the projection waveform exceeds the threshold values 31 and 32 used when the defect type is a point. get. As a result, the area included in the point defect can be acquired.

  The middle diagram of FIG. 22 shows a region including the region acquired in step ST702. This indicates the mask range determination process 3 in step ST703 executed by the concealment unit 1515. That is, the mask range determined by the concealment unit 1515 is a range obtained by expanding the area extracted in step ST702 shown in the upper diagram of FIG. 22 by a predetermined width in the main scanning direction and the sub scanning direction. ing.

  The lower part of FIG. 22 shows a defect image after masking the range determined by the masking unit 1515 by the mask range determination process 3. As a result of the point defect being concealed from the defect image information, the defect image information has no defect.

  The lower part of FIG. 22 represents a set determination feature value vector X3 having the set determination feature value calculated by the set determination value calculation unit 1511 as an element, as in the middle diagram of FIG.

The lower part of FIG. 22 represents that the length of the set determination feature value vector X3 does not exceed a predetermined threshold.
Therefore, the defect type determination process in FIG. 19 ends after determining two defect types, horizontal lines and dots.

  In the above embodiment, the center vector of the cluster is defined as the center vector of the smallest sphere that includes the coordinate group represented by the set determination feature value of the defect constituting the cluster. However, the present invention is not limited to this. For example, it is also possible to employ a configuration that is determined as an average value of coordinate groups represented by a set determination feature value of defects that constitute a cluster.

  In the embodiment described above, the calculation of the distance D1 between the cluster and the set determination feature value vector X has been described by calculating the Mahalanobis distance. However, the present invention is not limited to this. For example, the Euclidean distance, standard or Euclid A configuration in which the distance, the Euclidean square distance, the Manhattan distance, the Chebyshev distance, or the Minkowski distance is calculated can be employed.

  In the above embodiment, the length of the set determination feature value vector is calculated by calculating the length between the set determination feature value vector and the origin, and the distance calculation is performed by calculating the Mahalanobis distance. However, the present invention is not limited to this. For example, a configuration in which the Euclidean distance, the standard or Euclidean distance, the Euclidean square distance, the Manhattan distance, the Chebyshev distance, or the Minkowski distance is calculated can be employed.

1 is a configuration diagram illustrating an embodiment of an image forming apparatus of the present invention. It is a block diagram which shows the structural example of a failure diagnosis part. It is a figure for demonstrating the structural example of a defect classification determination part. It is the figure which showed notionally the structural example of the Bayesian network in the case of performing failure diagnosis of an image defect type | system | group. It is the figure which showed an example of the Bayesian network at the time of black line generation | occurrence | production in the structural example of the failure diagnosis by an image defect. It is a flowchart which shows an example of the control processing which a control part performs. It is a flowchart showing an example of the defect kind discrimination | determination process which a control part performs. It is a flowchart showing an example of the set determination feature value calculation process which a set determination value calculation part performs. It is a figure for demonstrating the relationship between a high gradation value waveform width and a threshold value. It is a figure for demonstrating the relationship between a cluster and a defect classification set. It is a figure for demonstrating the relationship between a set determination feature value vector and a cluster. It is a flowchart for demonstrating the primary stratification process which a set | judging determination part performs. It is a flowchart for demonstrating the secondary feature value calculation process which a classification determination value calculation part performs. It is a figure for demonstrating an example of the relationship between the defect image information whose defect classification is a horizontal line, and spectrum intensity. It is a figure for demonstrating an example of the relationship between the defect image information and defect intensity | strength whose defect classification is a finely divided fog. It is a figure for demonstrating an example of the relationship between the defect image information whose defect classification is a solid fog, and spectrum intensity. It is a figure for demonstrating an example of the relationship between the defect image whose defect classification is a point or a vertical line, and a projection waveform. It is a flowchart for demonstrating the secondary stratification process which a classification determination part performs. It is a flowchart showing the other example of the defect kind discrimination | determination process which a control part performs. It is a flowchart for demonstrating an example of the defect position specific process which a concealment part performs. It is a part of figure for demonstrating the defect classification determination process with respect to the defect image information which has two defect classifications of a point and a horizontal line. It is the other part of the figure for demonstrating the defect classification determination process with respect to the defect image information which has two defect classifications of a point and a horizontal line.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Image forming apparatus 110 ... Image acquisition part 120 ... Print engine part 130 ... Sensor part 140 ... Fault diagnosis information input part 150 ... Fault diagnosis part 160 ... Control part 1510 ... Defect type determination part 1511 ... Set determination value calculation part (set Judgment value calculation means)
1512 ... Set determination unit (set determination means)
1513 ... Type determination value calculation unit (type determination value calculation means)
1514 ... Type determination unit (type determination means)
1515 ... Concealment part (concealment means)
1520... Feature quantity extraction unit (feature quantity extraction means)
1530 ... Internal information acquisition unit (internal state information acquisition means)
1531 ... Component state information acquisition unit 1532 ... History information acquisition unit 1533 ... Environmental information acquisition unit 1540 ... Failure probability inference unit 1541 ... Failure candidate detection unit 1542 ... Inference engine 1543 ... Diagnostic model 1550 ... Additional operation information acquisition unit 1560 ... Diagnosis result Notification unit CH ... Horizontal line CHD ... Horizontal line / point CV ... Vertical line CVD ... Vertical line / point CVHO ... Vertical line / horizontal line / fogging DI ... Defective image information DM ... Main scanning direction DC ... Original DS ... Sub-main scanning direction HV ... Vertical Line I ... Integral value N ... High gradation value pixel count O ... Origin point TB ... Feature value table TS ... Defect type set table W ... Projected waveform X1-3 ... Plot by feature value of inspection image

Claims (8)

From the defect image information is information obtained from an image having its own output defect, calculated current Gotoku symptoms value for specifying a defect type group to the defect type is a type of a defect possessed by the defective image information element First calculating means for
Wherein on the basis of the defect type collector Gotoku symptoms value calculated in advance by the calculated center feature value and the first calculating means for each defect type set as the center of the set, and the set characteristic value and said central feature value A first specifying unit that calculates a distance for each defect type set, and specifies the defect type set corresponding to a central feature value used for the smallest of the calculated distances ;
From the defect image information, a second calculating means for calculating a seed-specific feature values for specifying the pre Kiketsu Recessed type,
Based on the species specific feature value calculated by the second calculating means, by the corresponding individual specific conditions determination processing each identified defect type set by said first specifying means, defects included in the said defect image information A second specifying means for specifying the type;
A failure diagnosis system comprising: Said first calculation means from the defect image information in a Defect image information which each pixel is represented by multi-value, it calculates a current Gotoku symptom value,
It said second calculation means, fault diagnosis according to claim 1 in which each pixel a the defect image information from the defect image information expressed in binary, to calculate the species-specific feature values, it is characterized by system. Further comprising concealment means for concealing the defect type identified by the second identification means from the defect image information;
Said first calculating means recalculates the current Gotoku symptom value from the hidden defect image information in said concealing means,
It said first specifying means, re-identifying the defect type group based on the current Gotoku symptoms value recalculated by the first calculating means,
Said second calculating means recalculates the species-specific feature values from the defect image information concealed by said concealing means,
Said second specifying means, based on the species-specific feature values recalculated by newly specified defect type group and the second calculating means by said first specifying means, and re-identifying the defect type, it The fault diagnosis system according to claim 1, wherein: Said second specifying means, the length of the current Gotoku Chochi vector to re calculated current Gotoku symptoms value element is re-identifying the defect type only if it exceeds a predetermined threshold by said first calculation means The fault diagnosis system according to claim 3. A failure diagnosis means for diagnosing a failure of each component constituting the image forming apparatus by analyzing a failure diagnosis model that models a cause of a failure of the image forming apparatus;
Internal state information acquisition means for acquiring internal state information of the device input to the fault diagnosis model;
A feature quantity extraction means for extracting a feature quantity characterizing a defect with respect to the defect type identified by the second identification means;
The failure diagnosis unit specifies a cause of failure by analyzing a failure diagnosis model corresponding to the defect type specified by the second specifying unit, using the information on the feature quantity and the internal state information. The fault diagnosis system according to any one of claims 1 to 4. The converging Gotoku Chochi the integral value of the gradation value histogram, gradation value which is the sum high tone value waveform width of a waveform width gradation value exceeds the threshold value in the projection waveform, and the standard deviation of tone value histogram 2. The standard deviation of the gradation value projection waveform, the number of peaks of the gradation value projection waveform, the maximum value of the gradation value projection waveform, and the intensity of the frequency component of the gradation value projection waveform. 6. The failure diagnosis system according to any one of 5 above. The species-specific JP Chochi, the maximum value of the projection waveform, as well as of claims 1 to 6, characterized in that the standard deviation of the spectral intensity of the maximum value and the projection waveform spectral intensity in the specific frequency region of the projection waveform The failure diagnosis system according to any one of the above. From the defect image information is information obtained from an image having its own output defect, calculated current Gotoku symptoms value for specifying a defect type group to the defect type is a type of a defect possessed by the defective image information element A first calculating step to:
Wherein on the basis of the defect type collector Gotoku symptoms value calculated in advance by the calculated center feature values for each defect type group and the first calculation step as the center of the set, the minimum among said calculated distance A first specifying step of specifying the defect type set corresponding to the central feature value used for the object ;
From the defect image information, and a second calculation step of calculating a species-specific feature values for specifying the pre Kiketsu Recessed type,
Based on the calculated species-specific feature values by the second calculating step, by said first individual specific conditions determination processing respectively corresponding to the identified defect type set by a particular step, the defect having the said defect image information A second specifying step for specifying the type;
A failure diagnosis method comprising:

Images