JP3498130B2 - Fault diagnosis knowledge tree hierarchy apparatus and fault diagnosis knowledge tree hierarchy method - Google Patents

Description

Detailed Description of the Invention

[0001]

The present invention relates may obtain the fault case used for fault diagnosis, organize the fault diagnosis system device such as a suitable fault diagnosis knowledge tree hierarchy apparatus for use in the fault diagnosis based on what, and Fault diagnosis knowledge tree hierarchization
In particular, the present invention relates to a fault diagnosis knowledge tree hierarchizing apparatus capable of efficiently generating a more optimized fault diagnosis knowledge tree that can be effectively used for fault diagnosis while reducing the burden on the operator. And trouble diagnosis knowledge tree
-Regarding hierarchical methods .

[0002]

2. Description of the Related Art An expert system is manufactured based on the knowledge and know-how of experts in the relevant technical field. This expert system, on behalf of the expert, gives answers to the problems that the expert should be in charge of. In particular, the diagnostic expert system is a system specialized for diagnosis and provides a means for fault diagnosis.

An expert system construction tool is used to support the construction of this expert system by a computer. However, in the conventional expert system construction tool, when acquiring knowledge from an expert in the relevant technical field, a knowledge engineer who is familiar with AI (artificial intelligence) needs to interview the expert and systematize the knowledge. was there. this is,
This is because the knowledge and know-how possessed by experts are rarely organized and systematized, and the level of knowledge used to create a knowledge base varies depending on the expert. In the case of organizing and systematizing only experts in the relevant technical field without the intervention of a knowledge engineer, it was necessary for the experts to create a framework for organizing and systematizing. Even if a framework is created, if the target equipment of the system is different, the application will not work.

In order to solve such a problem, in Japanese Patent Laid-Open No. 5-134872, knowledge about a target of failure diagnosis, that is, a target model, which is input / edited in a knowledge storage unit by a target model editor of a knowledge editing unit. And store
The causal relationship generation knowledge input and edited by the causal relationship generation knowledge editor is stored, and the causal relationship generation knowledge is applied to the knowledge about the fault diagnosis target input and edited by the target model editor in the causal relationship generation unit. By doing so, a knowledge acquisition device that automatically generates a causal relationship of a diagnosis target, and a diagnosis-type expert system using the knowledge acquisition device have been proposed.

Further, in Japanese Patent Laid-Open No. 6-44075, when a specific causal relationship to be diagnosed is instructed by an expert or the like, a condition relating to the structure from the failure case of the instructed specific causal relationship and the result.・ Retrieve and present constraints,
Unnecessary parts of the presented conditions / constraints are deleted, and the conditions / constraints remaining as a result of the deletion are used as conditions / causes for causal relationship generation knowledge.
It has a causal relationship generation knowledge acquisition unit that is used as a constraint and acquires the causal relationship generation knowledge by generalizing the specific causal relationship so that the specified specific causal relationship is given to the conclusion part. A knowledge acquisition device and a diagnostic expert system using the same have been proposed.

However, in the method of generating and utilizing "causal relationship generation knowledge" by the method described in JP-A-5-134872 or JP-A-6-44075, "step of generating target model", "causal relationship" Unless the three steps of "generation of relationship generation knowledge" and "step of generating causal relationship" are performed, the final organized and systematic knowledge will not be generated, and a great burden will be imposed on the expert. In addition, when the knowledge is organized and systematized by a plurality of specialists, there is a possibility that terminology may become ununiform, resulting in inconsistent knowledge. Further, since the flow chart is not displayed when inputting, there is a problem that it is not easy to organize and systematize knowledge.

Therefore, in Japanese Patent Application No. 7-262647, which has not been published at the time of filing of the present invention, a person including the inventors of the present invention proposes a technique relating to a failure case acquisition / arrangement device and a knowledge acquisition type failure diagnosis system. is doing. In this technology, a failure causal relationship input unit for inputting a series of knowledge of the failure causal relationship from the step of pursuing the cause of the generated failure case to the failure recovery step, and the input term input by the failure causal relationship input unit Supports the refinement of the knowledge input by the fault-causal relationship input unit using a term dictionary management unit that has a term dictionary in which terms related to equipment subject to failure diagnosis are registered for unified management, and a flow diagram And a knowledge refining support section. Further, in this technique, for example, in the failure causal relationship input unit, the knowledge of causal relationship of “sign”, “behavior”, and “result” is input and arranged. Further, for example, the failure causal relationship input unit is a "sign" of the failure that has occurred.
The “behavior” performed according to this and the “result” caused by this behavior can be input in order in chronological order. Therefore, by using such a failure case acquisition / arrangement device, it is possible to easily acquire organized and systematic knowledge, and it becomes easy for an expert to acquire, organize and systemize knowledge.

[0008]

However, the knowledge organized and systematized in this way is acquired by such a failure case acquisition / arrangement device, and then the acquired knowledge is used for the failure that has been conventionally used. When used as a knowledge acquisition type failure diagnosis device in combination with a case diagnosis device, for example, a device as in Japanese Patent Application No. 3-214320, there is a problem in that it is efficiently processed. Japanese Patent Application No. 7-26264
Even if the organized and systematic knowledge can be easily acquired by the technology proposed in Section 7, the failure diagnosis knowledge tree that can be used more effectively in the failure diagnosis of the failure case diagnosis device that performs subsequent processing. There are many inadequate points in terms of obtaining. Therefore, in the process of optimizing the failure diagnosis knowledge tree in the "order for efficient diagnosis" (hereinafter referred to as failure diagnosis knowledge tree optimization work), a large burden is placed on the expert. There is.

Here, this fault diagnosis knowledge tree optimization work will be briefly described.

[0010] When dealing with a failure, an expert may proceed with an investigation and identify the cause while performing "action that affects the system". This "action that affects the system" is an action that affects the system that is the target of the failure diagnosis (hereinafter, referred to as a failure search action), by canceling the fault that has occurred, so as to proceed to the investigation of the next stage. ). Further, if a fault diagnosis knowledge tree is generated based on such a case, it becomes a fault diagnosis knowledge tree including a fault search action.

Considering the failure search behavior, the symptom of a failure before and the failure symptom after a certain failure search behavior should not be replaced beyond the failure search behavior. However, in a failure case diagnosis device that creates a failure diagnosis knowledge tree using a known technique such as ID3 that organizes knowledge so that the diagnosis path is shortened, in one created failure diagnosis knowledge tree, With some fault-finding behavior, the attributes used in many cases are concentrated at the top of the fault diagnostic knowledge tree. That is, in many cases, the failure diagnosis knowledge tree is created so that the attribute that is the inspection item to be checked after the failure search action is also checked before the failure search action. Therefore, in order to improve the processing efficiency of the failure case diagnosis device that performs the failure diagnosis, many constraint conditions are set in order to perform the failure diagnosis knowledge tree optimization work in which the failure diagnosis knowledge trees are aligned and optimized in chronological order. Must be entered by an expert. Therefore, there is a problem that a large amount of manual operation occurs until the optimum failure diagnosis knowledge tree is completed, and the load for generating the failure diagnosis knowledge tree increases.

The present invention has been made to solve the above-mentioned conventional problems, and provides a more optimized failure diagnosis knowledge tree that can be effectively used for failure diagnosis while reducing the burden on the operator. A fault diagnosis knowledge tree hierarchical device that can be generated more efficiently, and a fault diagnosis knowledge tree.
It is an object of the present invention to provide a method for layering Lee .

[0013]

According to the present invention, a large fault diagnosis knowledge tree having a larger number of knowledge is divided into a plurality of smaller fault diagnosis knowledge trees having a smaller number of knowledge, and they are linked to each other. In addition, by layering these large failure diagnosis knowledge tree and small failure diagnosis knowledge tree, and canceling the occurring failure, the next step
System impact on the failure diagnosis target system
A failure case is time
In each divided partial diagnostic knowledge unit
A fault diagnostic knowledge tree hierarchical setting device for generating a fault diagnostic knowledge tree and a fault diagnostic knowledge tree hierarchical display device for hierarchically displaying the large fault diagnostic knowledge tree and the small fault diagnostic knowledge tree are provided. By doing so, the above problems are solved.

Next, a large failure diagnosis with a larger number of knowledge
Break the knowledge tree into multiple smaller
It is divided into fault diagnosis knowledge trees and linked together.
These large fault diagnosis knowledge tree and small
Error diagnosis knowledge tree is hierarchized and
So that you can proceed to the next stage of investigation by releasing the breakdown
Failure detection line that affects the target system
The failure case is divided into
A fault diagnosis knowledge tree for each partial diagnosis knowledge unit
The above-mentioned problem is solved by the generation.

[0015]

[0016]

The operation of the present invention will be briefly described below.

FIG. 1 is a fault diagnosis knowledge tree according to the present invention.
It is a block diagram showing the basic composition of a hierarchization device .

[0019] As shown in FIG. 1, the late <br/> impaired diagnostic knowledge characteristics of the tree hierarchy apparatus 16 according to the present invention, the fault diagnosis knowledge tree hierarchical setting device 42, the fault diagnosis knowledge tree hierarchical view It is basically configured by the device 36. Further, the fault diagnosis knowledge tree hierarchizing device 16 may be provided with a storage device 20 for storing various related files and knowledge data in addition to these.

First, the fault diagnosis knowledge tree hierarchy setting device 42 divides a large fault diagnosis knowledge tree having a larger number of knowledge into a plurality of fault diagnosis knowledge trees having a smaller number of knowledge, and links them in association with each other. . The fault diagnosis knowledge tree hierarchy setting device 42 hierarchically divides and links the large fault diagnosis knowledge tree and the small fault diagnosis knowledge tree.

The fault diagnosis knowledge tree hierarchical display device 3 is also provided.
Reference numeral 6 hierarchically displays the large failure diagnosis knowledge tree and the small failure diagnosis knowledge tree obtained by the failure diagnosis knowledge tree hierarchy setting device 42. The fault diagnosis knowledge tree hierarchical display device 36 can display the fault diagnosis knowledge tree in a hierarchical manner as described above, so that the display range of the fault diagnosis knowledge tree can be set more effectively.

For example, when the failure diagnosis knowledge tree to be displayed is too large to be displayed, the display range can be easily set according to the hierarchy of the failure diagnosis knowledge tree. Alternatively, for example, when the failure diagnosis knowledge tree to be displayed is too large to be displayed, it is possible to specify the size and the display range of the failure diagnosis knowledge tree.

According to the present invention having such a configuration, it is possible to more efficiently generate a more optimized fault diagnosis knowledge tree that can be effectively used for fault diagnosis while reducing the burden on the operator. it can.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 2 shows a first fault diagnosis system device having a fault diagnosis knowledge tree hierarchy device to which the present invention is applied.
It is a block diagram which shows the structure of embodiment.

In this embodiment, as shown in FIG.
In addition to the failure diagnosis knowledge tree hierarchy device 16 to which the present invention is applied, a failure case acquisition / arrangement device 12 and a failure case diagnosis device 14 are provided. The fault diagnosis knowledge tree hierarchization device 16 has basically the same configuration as that of FIG. 1 described above, and in addition to the fault diagnosis knowledge tree hierarchization setting device 42 and the fault diagnosis knowledge tree hierarchization display device 36, a storage device. Two
It has 0. In addition, the failure case acquisition / arrangement device 12
The failure-causal relationship input device 33 is provided.

The basic idea of this embodiment is to perform a fault search action in order to efficiently generate an optimized fault diagnosis knowledge tree that can be effectively used for fault diagnosis while reducing the burden on the operator. The failure cases are temporally divided as a unit of each, and each divided one is called a partial diagnosis unit, and the failure diagnosis knowledge tree is generated or optimized by such a partial diagnosis unit. . One feature of this embodiment is that the concept of such a partial diagnosis knowledge unit is introduced. In this embodiment, by introducing the concept of generating a failure diagnosis knowledge tree in such a partial diagnosis knowledge unit, the order of the attributes forming each part divided in such a partial diagnosis knowledge unit is It doesn't matter. Therefore, the optimized fault diagnosis knowledge tree can be efficiently generated or optimized.

Note that the fault exploration action is the action of canceling the fault that has occurred, and by canceling in this way,
It is an action to proceed to the investigation of the cause of failure in the next stage. In the present embodiment and the second embodiment to be described later, the following items are targeted as such failure detection actions.

A1. Pressing the fault reset button. A2. Turning on / off the control power supply by operating a power breaker, etc., for replacing cards (printed circuit boards), detectors, transistors, etc., fuses, etc. A3. Restart the computer used for control. A4. Restart the sequencer used for control.

The present embodiment will be described with reference to FIG. 2. First, an expert searches for the individual fault exploration behavior acquired and rearranged by the fault causal relationship input device 33 provided in the fault case acquisition / rearrangement device 12, and the fault diagnosis is performed. The failure detection behavior is marked by the knowledge tree hierarchy setting device 42. The operation of this mark is performed, for example, by operating the mouse on the window displayed on the display device of the failure diagnosis knowledge tree hierarchy setting device 42. The fault diagnosis knowledge tree hierarchy setting device 42 stores the fault case data marked with the fault exploration behavior in this way by the storage device 20.
Save to. At the time of saving, the case is separated immediately before each mark position (line), and saved as small case data in divided partial units, that is, partial diagnostic knowledge units.

Further, the fault diagnosis knowledge tree hierarchical setting device 4
2 divides a large fault diagnosis knowledge tree having a larger number of knowledge into a plurality of small fault diagnosis knowledge trees having a smaller number of knowledge, and links them by associating them with each other so that the large fault diagnosis knowledge tree and the small fault knowledge tree can be linked. Hierarchical diagnosis knowledge tree. The fault diagnosis knowledge tree hierarchization setting device 42 links the fault diagnosis knowledge tree of a layer one level higher from the top node of the fault diagnosis knowledge tree. Further, the fault diagnosis knowledge tree hierarchy setting device 42
From the bottom node of the fault diagnosis knowledge tree, a link is set to another fault diagnosis knowledge tree one layer below. For the fault diagnosis knowledge tree in the top layer, no link is set from the top node of the fault diagnosis knowledge tree. If the bottom node of the failure diagnosis knowledge tree is the failure cause node, no link is set from the bottom node of the failure diagnosis knowledge tree to the failure diagnosis knowledge tree one level below.

Next, the failure case diagnosis device 14 reads a plurality of failure case data from the storage device 20. Further, the failure case diagnosis device 14 collectively displays the failure cases of the read failure case data on the display device of the failure case diagnosis device 14 in a table format in the unit of partial diagnosis knowledge.
For such a tabular display, an expert edits attributes, attribute values, classes and the like by operating the mouse.

After the editing is completed, the failure case diagnosis device 14
Based on the knowledge base generated in (3), the fault diagnosis knowledge tree hierarchical display device 36 is used to display the optimized fault diagnosis knowledge tree. The display is performed on the display device included in the fault diagnosis knowledge tree hierarchical display device 36.
In the display, the fault diagnosis knowledge tree is displayed in hierarchical units. For example, when the top node of the failure diagnosis knowledge tree is clicked, the failure diagnosis knowledge tree of the hierarchy one level higher is displayed on the display device. On the other hand, when the node at the bottom of a certain fault diagnosis knowledge tree is clicked, the fault diagnosis knowledge tree of the next lower layer is displayed on the display device.

When the size of one hierarchy exceeds the range that can be grasped by humans, the fault diagnosis knowledge tree hierarchy display device 3
It is possible to sequentially display the designated number of hierarchical levels by using 6. The designation of this display is made in the form of how many nodes from a certain node to the upper hierarchy, or how many nodes from a certain node to the lower hierarchy.

The fault diagnosis knowledge tree hierarchical setting device 42, the fault diagnostic knowledge tree hierarchical display device 36, and the fault diagnostic knowledge tree displayed on the display device of the fault case diagnostic device 14 are the ones expected by the expert. If not, the knowledge base stored in the storage device 20 is edited again using the failure diagnosis knowledge tree hierarchy setting device 42, the failure diagnosis knowledge tree hierarchy display device 36, or the failure case diagnosis device 14. On the other hand, if the failure diagnosis knowledge tree is as expected by the expert, the generation of such a knowledge base is terminated.

Here, in the present embodiment and the second embodiment described later, the apparatus to be subjected to the failure diagnosis (hereinafter referred to as the failure diagnosis target system) is a continuous rolling apparatus for rolling a strip-shaped steel sheet. There is. When a failure actually occurs, the expert in the failure diagnosis target system displays the diagnosis screen of the failure case diagnosis device 14 on the display device, and performs diagnosis while inputting to the screen. The screen of the failure diagnosis knowledge tree hierarchical display device 36 is called from the failure case diagnosis device 14 and displayed on the display device. On the screen of the fault diagnosis knowledge tree hierarchical display device 36, as described above, the fault diagnosis knowledge tree is displayed in hierarchical units. If the top node of the displayed fault diagnosis knowledge tree is clicked, the fault diagnosis knowledge tree one level higher is displayed.
On the other hand, if the bottom node of the displayed fault diagnosis knowledge tree is clicked, the fault diagnosis knowledge tree one level below is displayed.

In the present embodiment, the failure diagnosis knowledge tree hierarchy device 16 and the failure case acquisition / arrangement device 12 are used.
The failure case diagnosis device 14 is a CRT (cathode ray tu
It is configured on a single computer device having a display device using be) and an input device using a mouse or a keyboard. Therefore, the failure case acquisition / arrangement device 12
The display of each of the failure diagnosis knowledge tree hierarchy setting device 42, the failure diagnosis knowledge tree hierarchy display device 36, and the failure case diagnosis device 14 provided in the failure diagnosis knowledge tree hierarchy device 16 is displayed by a single display device. This is done in various windows. In addition, the fault diagnosis knowledge tree hierarchical setting device 4
Various settings made in 2 can be performed on the display by the fault diagnosis knowledge tree hierarchical display device 36.
That is, in the hierarchical display of the failure diagnosis knowledge tree performed by the failure diagnosis knowledge tree hierarchical display device 36, from the top node and the bottom node of the failure diagnosis knowledge tree to the failure diagnosis knowledge tree of another hierarchy. You can also set the link.

As described above, according to this embodiment,
A fault diagnosis system equipped with a fault diagnosis knowledge tree layering device that can efficiently generate a more optimized fault diagnosis knowledge tree that can be effectively used for fault diagnosis while reducing the burden on the operator. The excellent effect of being able to provide the device can be obtained.

Further, according to the present embodiment, as described above, by dividing the failure diagnosis knowledge tree by the partial diagnosis knowledge unit as a unit of the failure search action, the following excellent effects can be obtained. it can.

In many cases, the troubleshooting performed by the expert is performed by repeating the work of investigating the situation and the work of trying to recover from the malfunction. The task of recovering from this failure corresponds to the failure search action of the present invention. Here, by performing the failure search action of the work for recovering the failure, the failure is not always recovered, and in some cases, a clue for identifying the cause of the failure may be obtained.

Thus, depending on the failure case, the result of the failure search action may be the restoration of the failure or the clue for identifying the cause of the failure.
In some cases, the fault case data is a mixture of types of fault exploration behavior. In such a case, when a failure diagnosis knowledge tree is generated based on the failure case data, if one failure diagnosis knowledge tree is simply generated, the order of attributes is not always efficient.

This is because the information before the work for recovering the failure by the actual failure and the information after the operation are
Basically the order must be guaranteed. However, if an attempt is made to compensate for such an order, the failure diagnosis knowledge tree will become huge and the expert will not be able to grasp it. On the other hand, if such a sequence is not guaranteed and the fault diagnosis knowledge tree is generated by ignoring it, in many cases, the investigation that should be performed after the work to restore the fault is performed before the work. The fault diagnosis knowledge tree is generated in such an order.

Thus, based on the failure case data, simply 1
If one fault diagnosis knowledge tree is generated, the problem as described above occurs, but in the present embodiment, a fault case is divided into partial diagnosis units using the fault detection behavior as a unit, and each divided A failure diagnosis knowledge tree is generated for each partial diagnosis unit. By doing so, the above-mentioned problems can be solved or reduced. That is, it is possible to eliminate or reduce the problem that the failure diagnosis knowledge tree becomes huge and cannot be grasped by an expert if the order of the failure diagnosis knowledge tree is guaranteed. In addition, if the failure diagnosis knowledge tree is generated by ignoring the order, in many cases, the failure diagnosis knowledge tree is generated in an order such that the investigation that should be performed after the failure detection action is performed before the failure detection action is performed. It is possible to eliminate or reduce the problem of being rid of.

As described above, according to the present embodiment, failure cases are divided in units of failure search behavior, and a failure diagnosis knowledge tree is generated in each of the divided partial diagnosis knowledge units, thereby further optimizing. The failure diagnosis knowledge tree can be efficiently generated, and the burden on the operator can be suppressed.

Further, in addition to the effect obtained by dividing the fault diagnosis exploration action in units as described above, according to the present embodiment,
It is also possible to obtain the effect of improving workability on display as described below.

As the knowledge of fault investigation increases, the fault diagnosis knowledge tree grows larger and larger. When the failure diagnosis knowledge tree becomes large, it becomes difficult to overlook the failure diagnosis knowledge tree. For example, it becomes difficult to grasp the contents of the failure diagnosis knowledge tree when constructing or modifying the failure diagnosis knowledge tree. Alternatively, it becomes difficult to visually follow the failure diagnosis knowledge tree during diagnosis when a failure occurs in the failure diagnosis target system. In this embodiment, since the failure diagnosis knowledge tree is hierarchized, even if the failure diagnosis knowledge tree becomes large, it is easier to understand and trace it.

For example, when a fault diagnosis knowledge tree is displayed or traced, it is designated in the form of how many nodes upward from a certain node or how many nodes downward from a node.
As a result, the height of the failure diagnosis knowledge tree is reduced as a result, and the width is also decreased as a result, so that the expert can easily understand the failure diagnosis knowledge tree. As described above, in the present embodiment, it is possible to obtain an excellent effect in displaying the failure diagnosis knowledge tree.

FIG. 3 is a block diagram showing the configuration of the second embodiment of the failure diagnosis system device to which the present invention is applied.

As shown in FIG. 3, the failure diagnosis system apparatus 10 of the present embodiment comprises a failure case acquisition / arrangement apparatus 1
2, failure case diagnosis device 14, and failure diagnosis knowledge tree hierarchical display device 36. In addition to these, the failure diagnosis system device 10 includes a term dictionary 22 and a case information file 2 which are all configured on a predetermined storage device.
4 and a knowledge base file 26. Furthermore,
The failure diagnosis system device 10 includes a term reference function unit 55.
A case matrix format conversion unit 62, a diagnostic knowledge tree generation unit 64, and a diagnostic knowledge tree calling unit 66.

Further, the above-mentioned failure case acquisition / arrangement device 12
Is a term dictionary device 32 and a failure causal relationship input device 33.
And a knowledge refining support device 34. Furthermore, the failure case acquisition / arrangement device 12 uses the term registration function unit 52.
, Term reference function unit 53, and flow diagram conversion function unit 57
And a case input sheet conversion function unit 58.

On the other hand, the failure case diagnosis device 14 described above includes a knowledge base generation device 38 and a failure diagnosis device 39.

First, each component of the above-mentioned failure diagnosis system device 10 will be described below.

The term registration function unit 52 is used by the term dictionary device 32.
Is built into. The term dictionary device 32 conceptually displays as shown in FIG. This term registration function unit 5
In item 2, terms relating to the equipment subject to failure diagnosis are registered. The term dictionary device 32 registers a new term in the term dictionary by the term registration function unit 52 from the term used for input in the failure / cause relationship input device 33. This registration is performed by writing data from the term dictionary device 32 to the term dictionary 22. Also, the term dictionary device 32
Has the function of correcting and deleting already registered terms. To correct or delete this term, the term dictionary device 32
Is performed by deleting the data of the corresponding term in the term dictionary 22.

The term reference function unit 53 is used by the term dictionary device 32.
Is built into. When the term causal relationship input device 33 requests a term reference, the term dictionary device 32 reads the term referred to by the term reference function unit 53 from the term dictionary 22 and outputs it to the fault causal relationship input device 33.

Next, the failure / causal relationship input device 33 has a function of inputting a failure / causal relationship, and in addition,
It also has the function of a fault diagnosis knowledge tree hierarchy setting device to which the present invention is applied. The failure causal relationship input device 33
Conceptually, the display as shown in FIG. 5 is performed. First, regarding the function of inputting the cause-and-effect relationship, more specifically,
It has a function of organizing cases and a function of separating investigation and restoration. On the other hand, as a function of the fault diagnosis knowledge tree hierarchy setting device, the fault-causal relationship input device 33 is more specifically provided with a function of setting a fault exploration action to which the present invention is applied for a fault case, It has the function of hierarchical setting of fault diagnosis knowledge tree.

The fault-causal relationship input device 33 inputs the "signature" of the fault that has occurred, the "action (fault detection action)" performed in accordance therewith, and the "result" caused by the action in order of time series. Each of the “sign”, “behavior”, and “result” is shown in FIGS.
As will be described later using, target equipment name, target equipment state,
It is possible to enter the details of the activities. Figure 10
The entry fields for each item on the display screen of the failure-causal relationship input device 33 as shown in FIG. 12 are linked with the term dictionary device 32, and each entry field has a term reference function built in the term dictionary device 32. It is possible to select and use a term registered in the term dictionary 22 while using the section 53.

Next, the knowledge refining support device 34 conceptually displays as shown in FIG. The knowledge refinement support device 34 has a flow diagram conversion function unit 57 and a case input sheet conversion function unit 58 built-in, and a function of extracting necessary knowledge,
It has the function of deleting the behavior that worsens the failure and its result, the function of changing the diagnosis order, the function of drawing the line for investigation and recovery, and the function of setting the initial symptoms.

The knowledge refining support device 34 is a device for visually displaying, as a flow chart, a series of "symptoms", "actions", and "results" input by the failure-causal relationship input device 33. The knowledge refinement support device 34 allows the operator to confirm the displayed flow diagram and easily specify, change, or redisplay a necessary part, an unnecessary part, or a part whose order should be changed. It is a device equipped with a user interface function. The knowledge refining support device 34 is used as appropriate and is not necessarily required.

Next, the fault diagnosis knowledge tree hierarchical display device 36 to which the present invention is applied hierarchically displays the fault diagnosis knowledge tree. Fault diagnosis knowledge tree hierarchical display device 3
6 conceptually provides a display as shown in FIG.

In the present embodiment, a large fault diagnosis knowledge tree having a larger number of knowledge pieces is divided into a plurality of small fault diagnosis knowledge trees having a smaller number of knowledge pieces, and the small fault diagnosis knowledge tree pieces are linked to be linked to each other. The knowledge tree and the small fault diagnosis knowledge tree are hierarchized. In particular, such hierarchization is performed by the failure / cause relationship input device 33 described above.

Here, the fault diagnosis knowledge tree hierarchical display device 36 hierarchically displays the large fault diagnosis knowledge tree and the small fault diagnosis knowledge tree. In such a display, the failure diagnosis knowledge tree hierarchical display device 36
Uses the diagnostic knowledge tree generation unit 64 and the diagnostic knowledge tree calling unit 66 incorporated in the fault diagnostic knowledge tree hierarchical display device 36.

The knowledge base generation device 38 has a function of integrating cases and a function of generating a failure diagnosis knowledge tree. The knowledge base generation device 38 conceptually displays as shown in FIG. The knowledge base generator 38
In such a function, the data of the term dictionary 22 is referred to using the term reference function unit 55 and the term dictionary device 32, and the data of the case information file 24 or the knowledge base file 26 is passed via the case matrix format conversion unit 62. Refer to. The knowledge base generation device 38 has a function of integrating cases and a function of generating a failure diagnosis knowledge tree.

The failure diagnosis device 39 is a failure causal relationship input device 3 having the function of a failure diagnosis knowledge tree hierarchy setting device.
In addition to having a function of displaying the failure diagnosis knowledge tree in hierarchical units set in 3, the display has a function of displaying the specified number of stages (number of layers) of the failure diagnosis knowledge tree. The failure diagnosis device 39 conceptually mainly displays as shown in FIG. For such a function, the failure diagnosis device 39 reads the data of the term dictionary 22 and the data of the knowledge base file 26 via the term reference function unit 55 and the term dictionary device 32.

10 to 12 are diagrams showing screens displayed on the display device of the failure / causal relationship input device 33 shown in FIG. These FIGS. 10 to 12 originally show one screen, and are divided into three for convenience of drawing.
10 shows the left side of the screen, FIG. 11 shows the center of the screen, and FIG. 12 shows the right side of the screen. The failure causal relationship input device 33 has the function of the failure causal relationship input device 33 related to the input of the original causal relationship and the function of the failure diagnosis knowledge tree hierarchy setting device. The "effects on the system" in Fig. 3 is only the part relating to the function of the fault diagnosis knowledge tree hierarchy setting device, and the other is the part relating to the original function of the fault / causal relationship input device 33.

"Fault name" and "Fault cause" at the top of FIG. 10, "Facilities" at the top of FIG. 11, and "Major items" from FIG. 10 to FIG. 11 in the table portion. Regarding the “Signs” item and the “Initial Signs” item of the “Results” of the large items from FIG. 11 to FIG. 12 in the table part,
Used to categorize failure cases to generate failure case groups. This failure case group corresponds to each knowledge base stored in the knowledge base file 26, which is used when the failure case diagnosis device 14 makes a diagnosis.

The item "restoration start position" at the leftmost end of the table shows the point at which restoration starts for one failure case. The rows above the row in which the recovery start point is set in the tables of FIGS. 10 to 12 are the knowledge that constitutes the failure diagnosis knowledge tree.

Next, the second item "necessary / unnecessary" from the left end of the table in FIG. 10 is an item for checking whether the contents of the relevant line are knowledge necessary for failure diagnosis or not. is there.

The third "influence on the system" from the left end of the table in FIG. 10 is the fault diagnosis knowledge tree hierarchical setting device, which is the action in which the content of the action in the relevant line affects the system. Therefore, this is an item for checking whether or not the failure detection behavior of the present invention or such behavior. The contents above the line set as the fault detection action marked with a circle in this "Impact on the system" and the contents on the lower side including the line which is the fault detection action are different faults. It also has a role of dividing the information into a diagnostic knowledge tree.

The large item "signature" in the table portion from FIG. 10 to FIG. 11 represents the sign that triggered the large item "behavior" in the table portion of FIG. Here, in the "signs" of the major item, "equipment / machine" indicates an equipment or machine in which a failure has occurred, "location" indicates the installation location of such equipment or machine, and "attribute name". Indicates the content of the failure, the "attribute value" indicates the classification of the failure, and the "incidental condition" indicates various conditions related to the failure. Here, the input criteria using the screens shown in FIGS. 10 to 12 is that the “attribute name” of “equipment / machine” in “location” is “attribute value” under “incidental conditions”. That is the case. In addition to these sub-items,
There are items "initial sign", "sign (new)", "sign (old)", and "memo".

Next, the large item "behavior" shown in the table portion of FIG. 11 represents the action taken in response to the above-mentioned large item "signature" as a whole. Here, to explain the small items in the "action" of this major item, the "how to do" item indicates the action content, the "equipment / machine" item indicates the equipment or machine that performed the action, and the "place" The item of indicates the installation place of such equipment or machine, the item of "for what" indicates what part of the equipment or machine the action taken is,
The item “timing” indicates the timing of the action taken. Therefore, the criteria for inputting "behavior" of the major items are "for what" of "equipment / machine" in "place",
It is the action of "what to do" at the timing indicated by "timing". In addition to the above-mentioned small items, there are items "number (new)" and "number (old)".

The structure and use of the small items of the large item "result" in the tables of FIGS. 11 to 12 are as shown in FIG.
This is the same as the configuration and usage of the small items of the "signs" of the large items in the table portion from FIG.

Here, the failure causal relationship input device 33 inputs the large items and the small items as described above by selecting the displayed terms. The displayed terms are the term dictionary 2
What is stored in No. 2 is read by the failure causal relationship input device 33 via the term dictionary device 32 and the term reference function unit 53, and displayed as a selection menu.

Here, the rules for reading the "signs", "actions", and "results" of the above-mentioned major items in FIGS. 10 to 12 are as follows.

Rule A1: Read from left to right, top to bottom. Rule A2: When the "behavior" for the "symptom" is not entered in the same line, the "sign" is treated in parallel with the "sign" in the next line. Rule A3: When the "result" for the "action" is not entered in the same line, the "action" is treated in parallel with the "action" in the next line. Rule A4: When the "action" for the "result" is not entered in the same line, the "result" is treated in parallel with the "result" in the next line. Rule A5: When the "symptom" for the "behavior" is not entered in the same line, the "behavior" is treated in parallel with the "behavior" in the next line.

In accordance with the above rules A1 to A5, FIGS.
Reading what is shown as an example in FIG. 12 is as follows.

Content B1: "5BR1" is displayed on the failure monitoring CRT.
Roll INV transmission error "was displayed. Content B2: In response to the content B1 described above, the failure reset button on the operation panel was pressed. Content B3: As a result of the content B2 described above, "transmission abnormality" is displayed on the failure monitoring CRT. Content B4: In response to the content B3 described above, the "INV board panel display" on the INV board was checked. Content B5: As a result of the content B4 above, "8.8" lights up in "INV board panel display" (CPU clock error) 21 Hz is displayed in "INV board panel display" and "OFF" in "INV board panel display" Lighting (INV power supply trip) “FAULT” is lighting on the “INV panel display” (IN
V breakdown) "STOP" lights up on the "INV panel display" (CPU
STOP) Content B6: In response to all the results of the content B5 described above, the control power supply is turned off and then turned on. Content B7: As a result of the content B6 described above, "RUN" lights up in "INV panel display" (normalization of INV control system)

Here, the case data generated by the failure causal relationship input device 33 described with reference to FIGS. 10 to 12 is not directly passed to the failure case diagnosis device 14, but is provided in the failure causal relationship input device 33. It is divided at the time of saving by the function of the fault diagnosis knowledge tree hierarchical setting device.
This division is based on the action for which the small item “impact on the system” in FIG. 10 is set. For example, in FIG. 10 to FIG. 12, the action (reset of serious failure) in which the small item “number (new)” of the large item “action” of FIG. 11 is “1” is set to “impact on system”. It is divided by being. or,
"Number (new)" action is "3" (control power "off" →
For "ON" operation, the small item "Impact on system" is set, but the small item "Recovery start position" is set for the same line, so "Sign" and "Result" above the same line are set. To save. In addition, the contents of the "result" column in which the small item "symbol (new)" of the large item "result" of FIG. 12 is "D" is set to "necessary / unnecessary" in the row of the result column. Therefore, the contents in the "Result" column are not saved. less than,
The contents of the case data saved after division are shown.

[0078] ○ Case data for the fault diagnosis knowledge tree of the first layer INV transmission error ……………… Display ○ Case data for the second layer fault diagnosis knowledge tree INV transmission error ……………… Display “8.8” …………………… Lit “OFF” ………………………… Lit “FAULT” ……………… Lit “STOP” …………………… Lit

The respective case data described above with reference to FIGS. 10 to 12 are stored in the case information file 24. In addition, the knowledge base is generated in the knowledge base file 26 as a matrix table as shown in FIG. 13 together with other case data. Note that FIG. 13 is also a display screen of the knowledge base generation device 38.

A fault diagnosis knowledge tree as shown in FIG. 14 is generated based on the knowledge base thus generated. This FIG. 14 is displayed on the fault diagnosis knowledge tree hierarchical display device 36. Here, the failure case diagnosis device 14 used in the examples of FIGS.
D3, which is a knowledge organization algorithm for generating a failure diagnosis knowledge tree, is assumed.

The left side of FIG. 14, that is, the "first layer" indicated by the symbol W1 is as shown in detail in FIG. The upper right part of FIG. 14, that is, the part indicated by the symbol W2 is as shown in FIG. The part shown in the lower right part of FIG. 14, that is, the part shown by reference numeral W3 is as shown in detail in FIG. Here, the correspondence relationship between the items of FIG. 13 and the items of FIG. 14 (that is also the items of FIGS. 15 to 17) is as follows.

C1: "equipment / machine" + "attribute name" of "symptom (or result)" in FIGS. 10 to 12 ... "attribute" in FIG. 13 C2: "symptom (or result) in FIGS. ) ”“ Attribute value ”... Intersection C3 of“ attribute ”and“ class ”in FIG. 13:“ Cause of failure ”in FIGS. 10 to 12 ...“ Class ”in FIG.

Some of the attributes shown in FIG. 13 are not included in the case data saved in FIGS. 10 to 12. For example, the first layer INV board “Hd”, INV board “LINK”, IN
V board "ERR", INV board "dL", INV board "C"
D ", INV board" OL ", second layer-2 INV board" d "
L "and INV board" CD ". These attributes are the attributes included in other case data. The attribute value fields of these attributes are judged and filled by humans.

FIG. 14 is an example in which the fault diagnosis knowledge tree generated based on FIG. 13 is displayed by using the fault diagnosis knowledge tree hierarchical display device 36. If the hierarchical setting is not made, the failure diagnosis knowledge tree cannot be generated from the contents of FIG. This is because there are classes (backup roll transmission error) having different attribute values for the same attribute in different layers.

At the time of diagnosis, diagnosis is carried out while checking the attributes of the failure diagnosis knowledge tree of FIG. 14 from above. In this way, the fault diagnosis system apparatus of the type using ID3 can successfully convert the case data.

Although the fault diagnosis system device of the type using ID3 is used in the present embodiment, basically, when the fault diagnosis system device requiring the cause of the fault, the attribute and the attribute value is used, the present invention is used. Is valid.

As described above, in the present embodiment, the fault diagnosis system apparatus can be configured while applying the present invention, and the more optimized fault diagnosis knowledge can be effectively used for the fault diagnosis. The tree can be generated more efficiently while reducing the burden on the operator.
Further, in the present embodiment, particularly, the continuous rolling mill is set as the failure diagnosis target system, and excellent failure diagnosis can be performed.

[0088]

As described above, according to the present invention, a more optimized fault diagnosis knowledge tree that can be effectively used for fault diagnosis is generated more efficiently while reducing the burden on the operator. It is possible to obtain an excellent effect that it is possible to provide a fault diagnosis knowledge tree hierarchizing apparatus and a fault diagnosis knowledge tree hierarchizing method that can be performed.
Further, such an apparatus or method can be used in a failure diagnosis system apparatus as in the above-described embodiment, and in this case, effective failure diagnosis can be performed.

[Brief description of drawings]

Block diagram showing the basic configuration of the fault diagnosis knowledge tree hierarchy apparatus according to the invention, FIG

FIG. 2 is a block diagram showing a configuration of a first embodiment of a failure diagnosis system apparatus including a failure diagnosis knowledge tree hierarchy device to which the present invention is applied.

FIG. 3 is a block diagram showing a configuration of a second embodiment of a failure diagnosis system device to which the present invention is applied.

FIG. 4 is a diagram showing a display screen of a term dictionary device used in the second embodiment.

FIG. 5 is a diagram showing a display screen of a failure-causal relationship input device used in the second embodiment.

FIG. 6 is a diagram showing a display screen of a knowledge refining support device used in the second embodiment.

FIG. 7 is a diagram showing a display screen of a fault diagnosis knowledge tree hierarchical display device used in the second embodiment.

FIG. 8 is a diagram showing a display screen of a knowledge base generation device used in the second embodiment.

FIG. 9 is a diagram showing a display screen of a failure diagnosis device used in the second embodiment.

FIG. 10 is a diagram showing an example screen (left side) displayed on the failure / causal relationship input device of the second embodiment.

FIG. 11 is a diagram showing an example screen (center) displayed on the failure / causal relationship input device of the second embodiment.

FIG. 12 is a diagram showing an example screen (right side) displayed on the failure / causal relationship input device of the second embodiment.

FIG. 13 is a diagram showing an example of a screen displayed by the knowledge base generation device used in the embodiment.

FIG. 14 is a diagram showing an example of a screen displayed on the fault diagnosis knowledge tree hierarchical display device used in the second embodiment.

FIG. 15 is a diagram showing details of a part (first layer) of a screen displayed on the fault diagnosis knowledge tree hierarchical display device.

FIG. 16 is a diagram showing details of a part (second layer-1) of a screen displayed on the fault diagnosis knowledge tree hierarchical display device.

FIG. 17 is a diagram showing details of a part of the screen (second layer-2) displayed on the hierarchical display device for fault diagnosis knowledge tree.

[Explanation of symbols]

10 ... Failure diagnosis system device 12: Failure case acquisition / arrangement device 14 ... Failure case diagnosis device 16 ... Fault diagnosis knowledge tree hierarchy device 20 ... Storage device 22 ... Term dictionary 24 ... Case information file 26 ... Knowledge Base File 32 ... Term dictionary device 33 ... Failure causal relationship input device 34 ... Knowledge refinement support device 36 ... Fault diagnosis knowledge tree hierarchical display device 38 ... Knowledge Base Generation Device 39 ... Failure diagnosis device 42 ... Fault diagnosis knowledge tree hierarchical setting device 52 ... Term registration function section 53 ... Term reference function section 55 ... Term reference function section 57 ... Flow diagram conversion function unit 58 ... Case input sheet conversion function unit 62 ... Case matrix format conversion unit 64 ... Diagnostic knowledge tree generator 66 ... Diagnostic knowledge tree calling unit

─────────────────────────────────────────────────── ─── Continuation of the front page (72) Keisuke Kadono Keisuke Kadano 2-3-3 Uchisaiwaicho, Chiyoda-ku, Tokyo Kawasaki Steel Co., Ltd. Tokyo headquarters (72) Inventor Manabu 3-3-3 Toyosu, Koto-ku, Tokyo Kawatetsu Information Systems Co., Ltd. (56) Reference JP-A-2-42535 (JP, A) JP-A-7-47400 (JP, A) JP-A-6-110690 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) G05B 23/00-23/02 G05B 13/02 G06N 5/04

Claims (2)

(57) [Claims] 1. A large fault diagnosis knowledge tree having a larger number of knowledge pieces is divided into a plurality of small fault diagnosis knowledge tree pieces having a smaller number of knowledge pieces and linked in association with each other so that the large fault diagnosis knowledge tree and Hierarchize a small fault diagnosis knowledge tree, and
By canceling it, we will proceed to the next stage of investigation,
Unit of failure exploration behavior that affects the failure diagnosis target system
, The failure case is divided in time, and
Generate a fault diagnosis knowledge tree with these partial diagnosis knowledge units
A fault diagnosis knowledge tree hierarchical setting device, and a fault diagnosis knowledge tree hierarchical display device for hierarchically displaying the large fault diagnostic knowledge tree and the small fault diagnostic knowledge tree. Knowledge tree hierarchy device. 2. A large fault diagnosis knowledge tool having a larger number of knowledges.
Lee has multiple small fault diagnostics with less knowledge.
Split into knowledge trees and associate and link
With these large fault diagnosis knowledge tree and small fault diagnosis
The hierarchical knowledge tree is hierarchized and the faults that have occurred are cleared to proceed to the next stage of investigation.
Affect the target system for failure diagnosis.
Breakdown of failure cases in units of failure detection behavior
Fault diagnosis with each divided partial diagnosis knowledge unit
Fault diagnosis knowledge tool characterized by generating knowledge tree
Lee hierarchy method.