JPH0658716B2 - Failure diagnosis device and failure diagnosis system - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a failure diagnosis device that supports maintenance work performed by a maintenance staff when various devices or equipment fail, and particularly to a candidate for maintenance. The present invention relates to a failure diagnosis device and a failure diagnosis system suitable for supporting maintenance work for complicated devices, equipment, etc., in which a variety of (fault candidates) are difficult to investigate.

[Prior Art] In a conventional failure diagnosis device, in order to identify a final cause of failure for a certain symptom, as described in, for example, Japanese Patent Laid-Open No. 62-4663, the cause and effect of each symptom and the cause thereof are identified. The "knowledge" indicating the relationship is indicated by a fault tree structure, and the cause of the fault is investigated by tracing the fault tree.

Also, "Reasoning from First Principles in Electronic Troubleshooting," Intu, J, Manmachine Studies 19, pp. 403-423 (1983) (Reasoning from fi
rst principles in electronic troubleshooting, In
In the conventional failure diagnosis device described in t.J.Man Machine Studies 19, pp403-423), based on the information on the function of the component to be diagnosed, the behavior when the component is assumed to be normal is estimated, This is collated with the actual behavior to pinpoint the failure location.

[Problems to be Solved by the Invention] Among the above-mentioned conventional techniques, in the former case, the more complicated and large-scale the device, equipment, etc. to be diagnosed, the larger the failure tree becomes, and the larger the cause. The number of times a fault tree is searched for investigation becomes extremely large. Furthermore,
There are various modes for failure, and when trying to cope with each mode with this conventional technique, "knowledge" of empirical causal relationship of the cause of the symptom of the diagnosis target name component is required for each mode. There is a problem that the amount of "knowledge" becomes enormous, and the number of searches for investigating the cause increases.

On the other hand, in the latter of the above-mentioned conventional techniques, if the diagnosis target is a simple one such as a logic circuit, it can be easily dealt with, but if the diagnosis target is a device including a complicated function such as a computer. , There is a problem that it is difficult to actually model that function and it is difficult to apply.

An object of the present invention is to provide a failure diagnosis device and a failure diagnosis system that can quickly and efficiently investigate the cause of a failure even when a large-scale device having a complicated function or the like is targeted for the diagnosis group.

[Means for Solving the Problem] The above object is to store knowledge of failure candidates corresponding to failure circumstances in a storage device, obtain a set of failure candidates that are considered to be the cause of this failure when a failure occurs, and Separately, test knowledge having information about the existence range of a failure cause candidate for which a certain test or measurement (hereinafter, referred to as test) result is obtained is stored in a storage device for each test, This is achieved by narrowing down the failure candidates from the set of failure candidates which is the existence range of this failure cause (fault candidate) obtained when a certain test is executed. When the diagnosis target has a hierarchical structure, it is further achieved by adding hierarchical level information to each of the above knowledge and narrowing down the failure candidates for each hierarchical level. Also, before actually conducting the test,
From the test knowledge of the test candidates, it is possible to narrow down the test candidates by judging in advance whether the test can narrow down the failure candidates based on the existence range of the failure candidates corresponding to the test result and the set of failure candidates. , Further achieved. Further, the failure diagnosis system can be obtained by connecting the failure diagnosis device and the diagnosis target with a data transmission network.

[Function] When a certain fault occurs, a set of fault candidates that are considered to be the cause of this fault is obtained, the test result is obtained by applying it to a diagnostic target of a certain test, and the test result is obtained from this set of fault candidates. The failure candidates are narrowed down by obtaining a set of candidates, and finally the failure candidate of 1 is obtained. Even if the diagnosis target has a complicated structure, it is possible to easily and promptly investigate the cause of the failure by sequentially repeating this procedure. In addition, by evaluating whether or not the failure candidates can be narrowed down by the test before executing the test, it is possible to avoid unnecessary execution of the test,
It is possible to speed up the failure diagnosis. Furthermore, by connecting the failure diagnosis device and the diagnosis target with a data transmission network, the diagnosis target at a remote place can be quickly diagnosed.

It should be noted that the expression “a set of failure candidates” or “a range of existence of a failure candidate” based on a certain test result is used, but it goes without saying that the failure candidates can be narrowed down by using a set having a relationship with them. Of course, that case is also included in the scope of the present invention. It goes without saying that, for example, a set obtained by removing the range in which no failure candidate exists from the set of failure candidates may be obtained.

[Embodiment] An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a configuration diagram of a failure diagnosis device according to an embodiment of the present invention. The failure diagnosis apparatus comprises a diagnosis apparatus main body 1 which is an electronic computer, storage devices 2 and 3, an input device 4 such as a keyboard and a mouse, and an output device 5 such as a display device. The storage device 2 stores the knowledge of failure candidates and the knowledge of test candidates, which will be described later, necessary for executing the diagnosis. The storage device 3 temporarily stores an intermediate diagnostic result, and stores a candidate for failure part (failure candidate) 31 and a candidate for test 32 to be executed. When information necessary for diagnosis is input from the input device 4 of this failure diagnosis device, the diagnosis device main body 1 investigates the cause of the failure while narrowing down the failure location and test candidates and outputs the diagnosis result, as will be described later in detail. Output to the device 5.

FIG. 2 shows an example of knowledge stored in the storage device 2. As shown in the figure, knowledge is expressed in a frame format. Then, as shown in FIG. 9A, failure candidates corresponding to each failure event occurring in the diagnosis target are stored as knowledge. As a result, a set of various failure candidates that should be considered first for each failure event can be read from the storage device 2.

The knowledge shown in FIG. 2B is stored in the storage device 2 as the knowledge of the test candidates. This knowledge is knowledge about a test that separates a required failure candidate from a set of failure candidates listed as failure candidates to be considered for a certain failure event. This knowledge includes the test outline, test methods,
How to determine the test results, information indicating the range of failure candidates for each of the determined results, and the evaluation index when selecting the test, such as the ease of performing the test or the state of the diagnosis target after the test is before the test. Information such as whether or not it will change compared to is added to the test name. Here, the method of determining the test result is described in the processing procedure. As a result, when this test result determination method is referred to in the course of the diagnostic processing, it is determined according to the processing procedure described, which of the previously considered results the test result corresponds to. It

Next, the diagnostic processing executed by the diagnostic device 1 using these knowledge will be described with reference to the flowchart shown in FIG.

When a failure occurs in the diagnosis target and the failure event name is input from the keyboard 4 (step 11), knowledge of the failure circumstances vs. the set of failure candidates stored in the storage device 2 (second
A failure candidate of the corresponding failure event is searched from the figure (a)), and a set of failure candidates obtained by the search is stored in the failure candidate storage unit 31 of the storage device 3 (step 12). In the next step 13, the test candidates that can be further divided into the first failure candidate set obtained in the previous step 12 from the knowledge about the test (FIG. 2 (b)) stored in the storage device 2 Only the selected test candidates are stored in the test candidate storage unit 32 of the storage device 3 in the same manner as in the post-operation step 19. It is also possible to perform the processing of step 13 offline in advance and store the processing result, that is, the correspondence information between the failure event and the initial test candidate set in the storage device 2 or 3. In this case, step 13 is a process of only searching for a test candidate corresponding to the input event, and thus has an effect of reducing the processing time.

The next processing step 14 is processing for selecting a test to be executed from the test candidate set obtained in step 13. For example, the evaluation criteria for test selection as shown in FIG. 4 are prepared, and according to this criterion, the priority is based on the information about the index related to the test selection in the knowledge about the above-mentioned test (FIG. 2 (b)). And select the test candidate with the highest priority. At this time, reference should also be made to evaluation criteria such as ease of test execution and whether or not a disturbance that changes the state of the diagnosis target after the test is weakened. The outline and the test method of the test candidates thus selected are displayed on the display device 5 (step 15).

The diagnostician, that is, the maintenance factor, executes the test according to the instructed test procedure while observing this display. Then, when the diagnostician inputs the data obtained as a result of the test from the input device 4 (step 16), or when the diagnostic device 1 can automatically read the data of the test result, the data reading is executed in this step 16. And proceed to the next step 17.

In step 17, the input test result data is analyzed according to the processing procedure of the test result determination method that accompanies the knowledge of the test, and which test result is corresponded to is determined.
From the knowledge of the test shown in (b), information on the existence range of the failure location corresponding to the test result is extracted.

After extracting the information indicating the existence range of the failure part in step 17, the failure candidates are narrowed down from the failure part existence range as described later in detail (step 18), and this will also be described in detail later. In addition, the test candidates that can further narrow down the set of new failure candidates obtained in step 18 are narrowed down from the set of test candidates before the test (step 19
(Details will be described in FIG. 6). And step 20
Then, it is judged whether only one failure candidate remains or there are no test candidates, and if there are a plurality of failure candidates or there are test candidates to be executed, the process returns to step 14. When steps 14 to 20 are repeated and one failure candidate is narrowed down, the process proceeds from step 20 to step 21,
The finally obtained failure candidate is output to the output device 5, and this diagnosis processing is ended.

FIG. 5 is a detailed explanatory diagram of the failure candidate narrowing-down process in step 18 of FIG. Assuming that the failure candidate set before the execution of the test Ti is A and the range of the failures (the set of failure candidates) obtained by executing the test Ti is B, this set A, the set B, and the common overlap A∩ Find B = C,
This set C is set as a new set of failure candidates after the execution of the test Ti.

FIG. 6 is a detailed explanatory diagram of the test candidate narrowing processing in step 19 of FIG. Let D be the set of test candidates before the execution of test Ti, and let Tm (m = 1, 2,
, K) will be described below.

In the present embodiment, before actually performing the test, the evaluation in each test is simulated by the following processing to eliminate in advance a useless test in which failure candidates cannot be narrowed down.

First, in the case of evaluating a certain test Tm among the test candidates, the knowledge of the test candidate Tm (FIG. 2 (b)) is referred to and the range where the failure candidate exists for each test result j is extracted.
Then, as in the description of FIG. 5, a common set C of the failure candidate set C before the execution of the test Ti and the existence range of the failure candidate when the test result is j when the test Tm3 is executed.
Find j ′. Then, depending on the relationship between the sets C and Cj ', the following processing is performed to narrow down the test candidates.

1) Each set Cj 'for all test results j of test Tm
Is not an empty set, and the set Cj 'for some test result j is a true subset of the set C; in this case, this test Tm can narrow the set C, so this test The candidate Tm is a member of the new test candidate set E after the execution of the test Ti.

2) If the set Cj 'is an empty set for some test result j and the set Cj' is a true subset of the set C for another test result j; Become a member. However, when the process returns to step 14, this test Tm is selected, and when it is actually executed, the obtained set Cj 'may be an empty set depending on the test result j. When the set Cj 'after the test is actually an empty set, there are no fault candidates, and the result is inconsistent with the diagnosis result up to now. At this time, the diagnosis is proceeded by ignoring this test result, but this test Tm is recorded as a contradiction test. This test candidate T _m is displayed as a retest target, for example, when the diagnosis result at the end of diagnosis is incorrect.

3) For all test results j, each set Cj 'is Cj'
= C; In this case, the test candidate Tm cannot separate the failure candidate set C after execution of this test Tm, so the set E
Is not a member of

4) For all test results j, each set Cj 'is Cj'
= When it is an empty set; That is, this test Tm completely contradicts the test results up to now. In this case, the test candidate Tm is not a member of the set E. However, record Tm as a conflicting test.

5) For all test results j, each set Cj 'is Cj'
= When it is an empty set or C; In this case, the test Tm is not a member of the set E.
However, because the test candidate Tm is a partially contradictory test,
Record test candidate Tm in the record.

As described above, the knowledge information and the processing procedure are used to narrow down the failure candidates and the test candidates to be executed next are also sequentially narrowed down according to the failure candidates, so that unnecessary test execution can be avoided and the test to be executed can be avoided. Since knowledge of can be limited, appropriate diagnosis can be efficiently performed.

In the test candidate selection process in step 14 of the flowchart of FIG. 3, the test to be executed is selected according to the priority of the test candidate, but the present invention is not limited to this selection method. FIG. 7 shows another embodiment of the test candidate selection processing. In this embodiment, various test selection criteria are prepared, and the diagnostician selects appropriate selection criteria.

In the case of following this processing means, first, in step 101, the selection branch of whether to select the test candidates automatically or manually and the selection branches of various selection criteria when the automatic selection is selected are displayed on the display device. (One of the output devices 5 in FIG. 1). While observing this display, the diagnostician selects a selection branch using the input device 4 (step 102). If "automatic" is selected in this selection, the selection criteria to be applied are also selected at the same time.

In the next step 103, the diagnostician is "automatic" in step 102.
It is determined whether or not has been selected. If the determination result is negative, that is, if the diagnostician selects "automatic", the process proceeds to step 104. In this step 104, each test name of the test candidate set obtained in step 13 or step 19 of FIG. 3 and its test outline are displayed on the display device. While observing this display, the diagnostician can input an appropriate test from the test candidates using the input device 4
Specify with (step 105).

In the next step 106, what happens to the failure candidate when the designated test is temporarily executed is determined in the same manner as described with reference to FIG. 6, and this is displayed for each test result of the test. Since the evaluation of the test by this simulation has already been evaluated in step 13 or step 19 in FIG. 3, the evaluation result may be read out from the storage device and displayed.

When the determination result in step 103 is affirmative, that is, when “automatic” is selected in step 102, the process proceeds to step 107. In step 107, the highest priority test candidate is selected from the test candidates from the relationship shown in FIG. 8 and the selected test candidate is displayed. In FIG. 8, each test candidate is shown as a point in the space having the coordinates described in FIG. 2 (b) as coordinates, and the test closest to the origin is selected as the highest priority test. still,
In this embodiment, the closer to the origin, the higher the priority is. However, it goes without saying that the coordinates can be determined so that the farther from the origin, the higher the priority becomes. Further, FIG. 8 means a space in which n coordinate axes are set in correspondence with n selected indexes. As an index,
Various items such as easiness of test and establishment of failure can be prepared, but the values (coordinate values) of these indexes are set in advance for test knowledge and each failure candidate. In step 107 of FIG. 7, after displaying the highest priority test candidate, if there is only one test candidate, the process proceeds to step 106. If there are multiple test candidates, the process proceeds to step 105 to manually select an appropriate test candidate. . Of course, it is also possible to automatically decide the order of display.

According to this embodiment, the next test to be executed can be selected from the perspective of the diagnostician, and it is possible to know in advance what the failure candidates will be if the test is executed. The person can play a central role in the diagnosis.

FIG. 9 is a flow chart showing a procedure for executing the diagnosis shown in FIG. 3 stepwise by using the knowledge in which the components to be diagnosed are hierarchized. FIG. 10 shows the hierarchization of diagnosis targets.

In the diagnosis based on FIG. 9, the failure candidates and test candidates of each level are narrowed down from the upper level according to the hierarchy of FIG.
For this, the knowledge shown in Figure 11 is used. Fig. 11 (a)
2B and 2B are diagrams in which the diagnostic level information is added to the knowledge of the failure event-fault candidate and the knowledge of the test shown in FIGS. 2A and 2B, respectively. FIG. 11 (c) is knowledge about the hierarchical structure of the diagnosis target, and shows the lower-level components and the test items for all the lower-level components (fault candidates) for each component of the diagnosis target shown in FIG. It is an addition of collective knowledge. If these knowledges are used, the processing from step 201 to step 205 shown in FIG. 9 is carried out by using only the knowledge of the same diagnosis level to proceed the diagnosis processing, and the processing procedure shown in FIG. It is the same as 11 to 20). What differs from the processing procedure shown in FIG. 3 is processing for lowering the diagnostic level. When it is determined in step 205 that the failure candidates and test candidates narrowed down by the current diagnostic level can no longer be narrowed down, the process proceeds to step 206, and it is determined whether the current diagnostic level is the lowest. This step
If the determination result in 206 is "lowest", the process proceeds to step 209 to display the result, and this processing procedure ends.

Decrease diagnostic level if not "lowest" Step 20
Proceed to 7. In step 207, a lower-order failure candidate is generated with reference to the knowledge shown in FIG. 11 (c), and similarly in the next step 208, a lower-order test candidate is generated and the process returns to step 204 to diagnose this level. Start.

According to this embodiment, first, it is possible to specify a failure point from a rough failure point to a detailed failure point,
In addition, since the knowledge to be used is also limited for each diagnosis level, efficient diagnosis is possible especially when the diagnosis target is large-scale.

FIG. 12 is a configuration diagram of an embodiment in which the diagnostic device shown in FIG. 1 is connected to a diagnostic target device via a data transmission network. In the figure, 1 to 5 are the same devices as the devices _denoted by the same reference numerals in FIG. 1, 6 is a data transmission network, and 7 ₁ to 7 _n are devices to be diagnosed. By forming this network, the diagnostic device can automatically collect the test data from the diagnostic target device. Therefore, if the automatic data collection via this data transmission network and the interactive data collection from the input device shown in FIG. 1 are combined, more efficient diagnosis can be performed.

In addition, in the above-mentioned embodiment, although the simulation is performed and the test candidates are narrowed down before the test, it goes without saying that the actual test may be performed without performing the simulation.

[Effect of the Invention] According to the present invention, it is possible to eliminate useless tests by simulation, select only appropriate tests, obtain test data thereof, and narrow down failure candidates. However, it becomes possible to quickly and efficiently investigate the cause of the failure.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of a fault diagnosis apparatus according to an embodiment of the present invention, FIG. 2 (a) is a knowledge explanatory diagram of failure events vs. fault candidates, and FIG. 2 (b) is a knowledge explanatory diagram of test candidates, FIG. 3 is a flow chart showing a failure diagnosis processing procedure, FIG. 4 is an explanatory view of evaluation criteria for test selection, FIG. 5 is an explanatory view of narrowing down failure candidates, FIG. 6 is an explanatory view of narrowing down test candidates, and FIG. FIG. 8 is a flow chart of another embodiment of the processing, FIG. 8 is an explanatory view of test evaluation coordinates used in the test selection processing of FIG. 7, and FIG. 9 is a flow chart showing a processing procedure in an embodiment in which diagnostic processing is hierarchically executed. Figure 10 shows the hierarchical structure of the diagnosis target,
11 (a), (b), and (c) are diagrams for explaining knowledge in the case of performing diagnostic processing in a hierarchical manner, and FIG. 12 is a configuration diagram of a system in which a failure diagnostic device and a device to be diagnosed are connected by a data transmission network. Is. 1 ... Diagnostic device, 2, 3 ... Storage device, 4 ... Input device, 5 ... Output device, 6 ... Data transmission network,
7: Target device for failure diagnosis.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Tetsuo Ito 1168 Moriyama-cho, Hitachi-shi, Ibaraki Energy Research Laboratory, Hiritsu Seisakusho Co., Ltd. (56) Reference JP-A-63-269207 (JP, A)

Claims (6)

[Claims] 1. Storage means for storing knowledge about a failure candidate associated with a failure event, information about a range of existence of the failure candidate corresponding to a test result, and knowledge of a test candidate having an evaluation criterion, and a failure event when a failure occurs. Means for obtaining a set of failure candidates corresponding to, a means for selecting a test satisfying a specified evaluation criterion from among the test candidates, a means for obtaining a test result obtained by executing the selected test, and the test result And a means for extracting the existence range of the failure candidate corresponding to the above from the storage means, and a means for narrowing down the failure candidate by obtaining a common set of the existence range of the failure candidate and the set of failure candidates. apparatus. 2. Knowledge relating to a fault candidate having hierarchy level information corresponding to a fault event and having hierarchical level information according to a hierarchical structure of a diagnosis target, information regarding an existence range of the fault candidate corresponding to a test result, an evaluation criterion, and a hierarchy of a diagnosis target. Memory means for storing knowledge of test candidates having hierarchy level information according to structure, means for obtaining a set of failure candidates at a certain hierarchy level corresponding to a failure event when a failure occurs, and designated from among the test candidates Means for selecting a test at a certain hierarchical level that satisfies the evaluation criteria, means for obtaining a test result obtained by executing the selected test, and means for extracting the existence range of failure candidates corresponding to the test result from the storage means And a means for narrowing down the failure candidates at the hierarchical level by finding a common set of the existence range of the failure candidates and the set of failure candidates, and the reason for narrowing down by this means. When there is only one candidate or when there are no more test candidates to be executed, the hierarchy level is lowered and the same procedure is repeated to narrow down the failure candidates to a more detailed hierarchy level. Failure diagnosis device. 3. The method according to claim 1 or 2, wherein, before actually executing each test of the test candidates, information regarding the existence range of the failure candidate corresponding to the test result of each test is retrieved from the storage means to detect the failure candidate. A failure diagnosis apparatus comprising means for narrowing down test candidates by excluding test candidates for which failure candidates cannot be narrowed down from an existing range and a set of failure candidates. 4. The fault diagnosis apparatus according to claim 1 or 2, further comprising means for automatically selecting a test satisfying a designated evaluation standard. 5. A means for displaying a menu of test candidates on a display device, a means for an operator to select and select a test candidate from the menu according to claim 3, and temporarily execute the selected test candidate. And a means for displaying failure candidates narrowed down for each test result. 6. A failure diagnosis system comprising the failure diagnosis device according to any one of claims 1 to 5 and a diagnosis target connected through a data transmission network.