JPH06149577A - Method and device for fault diagnosis - Google Patents

Abstract

PURPOSE:To efficiently diagnose faults caused by physical trouble of apparatus and communication lines, erroneous operation, trouble of a program, malfunction due to defective setting of parameter values, etc. CONSTITUTION:The method and the device for fault diagnosis which diagnose faults occurring in apparatus and communication lines are provided with a diagnostic knowledge base 11 consisting of a diagnostic tree 11a, where diagnosis data described correspondingly to fault states are hierarchically arranged as plural nodes, and are provided with a diagnosis node designating part 21 which designates a node in the diagnostic knowledge base most related to the fault against which a countermeasure should be taken, a check part 22 which checks contents corresponding to the designated node, and a check result discriminating part 23 which selects the node showing the most definite fault state from candidates of nodes showing fault states.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is used for fault diagnosis of a device in which a large number of devices are complicatedly related. Examples of a device to be diagnosed (device to be diagnosed) in which the present invention is used are a workstation, a communication network in which a large number of communication paths are connected, a manufacturing robot in which a plurality of manufacturing robot devices and a cutting device operate in association with each other. These include equipment, traffic control devices for road networks and railroad networks, information processing devices in which a plurality of memories and printer devices are connected in a complicated manner, and the like.

The present invention relates to a method and apparatus for preserving a knowledge base in a computer device and performing fault diagnosis by comparing the operation of a device under diagnosis with the knowledge base. The present invention provides a method for pursuing the cause of a malfunction due to a physical failure of the apparatus hardware, a failure due to an erroneous operation of a user or an operator, a defect of software for controlling the apparatus, a deficiency of setting parameters, and the like. Used as a device.

[0003]

2. Description of the Related Art Recent devices, machines, communication paths, etc. are becoming larger in scale, and are becoming more complicated or wider. When a failure occurs in them, the loss is large. Therefore, it is very important to efficiently diagnose the failure of the equipment or the communication path and perform the early recovery.

As a conventional fault diagnosis, a method of arranging diagnostic information in a hierarchical manner, analyzing from higher information, and selecting the most probable lower information is a patent application by the applicant of the present application.
3-316037 “Network fault diagnosis data description method”. For these, a method of sequentially diagnosing from the highest information of the diagnostic knowledge base is adopted.

According to this method, even if the failure state is known to some extent and the inspection results of some nodes can be predicted, the inspection is sequentially performed from the highest node.

Another method is Japanese Patent Laid-Open No. 1-1267.
There is one disclosed in Japanese Patent No. 61. In this method, a computer predicts a motion based on a structural model of a fault diagnosis target, and a component related to a constraint condition that leads to a result in which the prediction is inconsistent with a symptom is diagnosed as a fault cause candidate.

[0007]

According to the conventional technique as described above, in the former case, since the diagnosis is performed from the highest node of the diagnostic tree, the intermediate part which is not the highest node of the diagnostic tree is previously obtained. Even when the symptom of the node (intermediate node) is known, or even when the diagnosis can be started from the middle like a senior maintenance person, the diagnosis is started from the highest level and the diagnosis is performed to some extent. It will take a lot of maintenance time. Further, in the latter case, since the operation prediction by the computer is required, the skill level of the maintenance person is required and it takes time for the prediction, and in any case, the failure diagnosis cannot be performed efficiently. There's a problem.

The present invention solves such a problem by using a diagnostic knowledge base composed of a diagnostic tree that describes inspection contents used for diagnosing a fault, and starts diagnosis from the middle of the diagnostic knowledge base. It is an object of the present invention to provide a device that enables the diagnosis and can efficiently perform diagnosis without performing unnecessary inspection.

[0009]

SUMMARY OF THE INVENTION The first aspect of the present invention is to use diagnostic data, which describes contents for inspecting a fault condition in association with each fault condition, as a node so that the node has a tree structure. In a method of holding a diagnostic tree arranged hierarchically as a knowledge base and performing fault diagnosis while sequentially tracing the nodes from upper to lower according to the hierarchical relationship of the diagnostic tree, a node in the middle of the diagnostic tree Is specified, and the fault diagnosis is performed while tracing downward from the specified node as a starting point.

In a second aspect of the present invention, diagnostic data, which describes contents for inspecting a fault condition corresponding to each fault condition, is used as a node, and the nodes are hierarchically arranged so as to form a tree structure. The diagnostic tree is held as a knowledge base, and the data obtained from the operation of the device under diagnosis and the diagnostic data held in this knowledge base are sequentially traced from the upper node to the lower node according to the hierarchical relationship of the diagnostic tree. While collating, a failure diagnosis device having means for selecting a node indicating a probable failure state and means for displaying the selected node as a diagnosis result is used to operate a node in the middle of the diagnosis tree. A feature is that a means for designating is provided.

[0011]

Operation: The diagnostic data is provided with a diagnostic knowledge base in which diagnostic data in which the contents for inspecting the fault condition are described in association with each fault condition are hierarchically arranged as a plurality of nodes and configured as a diagnostic tree. By specifying the node in the diagnosis tree to start, the inspection contents described in each node are inspected from the nodes specified according to the hierarchical relationship of the diagnosis tree, and among the node candidates showing some failure states The node showing the most probable fault condition is selected from the inspection results of the individual nodes. By repeating such operations, the node in which the cause of the failure is described is finally selected and the failure diagnosis result is displayed.

As a result, the diagnosis can be started from any designated node in the diagnosis tree, and the failure can be efficiently diagnosed without performing unnecessary inspection.

[0013]

Embodiments of the present invention will now be described with reference to the drawings. 1 is a block diagram showing the configuration of an embodiment of the present invention, FIG.
FIG. 6 is a diagram showing a configuration of a diagnostic tree of a diagnostic knowledge base in the embodiment of the present invention.

In the embodiment of the present invention, the input device 3, the display device 4 and the storage device 1 are connected to the central processing unit 2, and the central processing unit 2 has physical failures, erroneous operations and programs that occur in equipment and communication paths. And a diagnostic device 20 for diagnosing a fault caused by a malfunction due to improper setting of parameter values. Further, diagnostic data described in the storage device 1 in association with the fault state is arranged hierarchically as a plurality of nodes. A diagnostic knowledge base 11 configured as a diagnostic tree 11a is provided, and a diagnostic node designating unit 21 for designating the diagnostic device 20 the node most relevant to the fault to be dealt with among the nodes indicating the fault state of the diagnostic knowledge base 11. And the inspection unit 22 that inspects according to the contents described in the node of the specified hierarchy, and the most probable fault condition among the nodes indicating the fault condition. Was selected based on the node to the inspection result of the inspection unit 22 shown includes a test result determination unit 23 to be displayed on the display unit 4.

The diagnostic tree 11a is constructed in a hierarchical structure as shown in FIG. 2, and the diagnosis is performed from the upper part to the lower part. An inspection item is described at each branch point (hereinafter referred to as a node), and the inspection unit 22 of the diagnostic device 20 performs an inspection and executes a diagnosis by referring to the inspection item. When the inspection unit 22 refers to the diagnosis tree 11a of the diagnosis knowledge base 11,
The node instructed by the diagnostic node instructing unit 21 is inspected.

At the start of diagnosis, the diagnostic node instructing unit 21 receives the node identification name designated by the operator from the input device 3, and sends the identification name to the inspection unit 22. The input device 3 may be a device such as a mouse or a touch pen using a keyboard or a display device.

The inspection unit 22 confirms the inspection contents of the node having the instructed node identification name described in the diagnostic tree 11a and executes the corresponding inspection. The inspection result is sent to the inspection result determination unit 23.

The inspection result judging unit 23 converts the inspection result received from the inspection unit 22 into a numerical value as a certainty factor indicating the degree of certainty that the failure state of the node is related to the current failure. . As a conversion method, a coefficient is given to each check of the node, and it is obtained by calculating the coefficient. As an example for obtaining the certainty factor, R1 ..., Ri,
, Rn (0 <Ri <1) is given, the coefficient value Ri is calculated if a result corresponding to the fault condition is obtained in the inspection, and if not obtained, this coefficient value is calculated. 0
The confidence factor Pk of the node K can be given by [outer 1] by calculating. Where Π (1-Rn) is (1-R
1) × (1-R2) × ... × (1-Ri) × ... × (1-R
n) is shown.

[0019]

[Outer 1] The inspection result judging unit 23 judges whether or not the obtained certainty factor is a value corresponding to the failure state, and sends the result to the diagnostic node instructing unit 21. As the judgment criterion, it can be judged whether the confidence factor P of the node is higher or lower than the reference value.

If the inspection result does not match, the inspection result judging section 23 sends the information indicating that the node identification name and the instructed node are inappropriate to the display device 4 in order to inform that effect. In the display device 4, the display device or the like notifies that the instructed diagnosis start node is not appropriate. If the inspection result is appropriate, the inspection result determination unit 23 sends the node identification name to the diagnostic node instruction unit 21.

In the diagnostic node designating section 21, the diagnostic tree 11a
Is checked, the lower node of the node identification name sent from the inspection result determination unit 23 is searched, and the node identification name is sent to the inspection unit 22. Generally, there are multiple lower nodes,
The inspection unit 22 sends each inspection result to the inspection result determination unit 23.

If there is no subordinate node, it is a node indicating the cause obtained by the node, and the diagnostic node instructing unit 21 notifies the display device 4 that it is a node indicating the node identification name and the cause, and the display device 4 The cause and recovery measures are notified by a display device or the like.

The inspection result judging unit 23 calculates the certainty factor of each node, and the node identification name of the node having the highest certainty factor among the nodes having the certainty factor exceeding the reference value is given to the diagnostic node instructing unit 21. Will be sent. By repeating this, the cause of the failure is clarified.

After calculating the certainty factor in the inspection result judging unit 23, if there is no node having the certainty factor exceeding the reference value,
The inspection result determination unit 23 sends a message to that effect to the display device 4, and the display device 4 displays that an appropriate inspection result has not been obtained.

The embodiments of the present invention will now be described in more detail.

FIG. 2 shows a part of a diagnostic tree which is diagnostic knowledge of a network including a device called a modem for modulating / demodulating a communication signal and a host computer including a general computer, a personal computer and a workstation. The diagnosis is gradually started from the start of the knowledge called ROOT described at the left end of the diagnosis knowledge.

In this case, the analog connected upper side of the node connected to the right side of the root is abnormal (the upstream line of the analog line is abnormal), the lower analog state is abnormal (the downstream line of the analog line is abnormal), ... Inspect each symptom such as a test (need to test the child modem), and start by isolating the fault range of where the fault is specific to the analog / digital line or device.

Further, if it is judged as analog OK (there is no abnormality in the analog line, the digital system is a problem),
Underneath it, suspect a failure such as a clock setting error (clock mismatch or setting error), speed error (transfer speed error), ..., CAR error (parameter error related to compensation operation at line disconnection). , Test and confirm about each. By repeating this, the cause of the appropriate failure is investigated.

According to the present invention, for example, when the maintenance staff does not set the clock in the newly installed modem, or when he / she notices that the parameter different from the other is input at the time of starting the diagnostic system, The diagnosis can be started by instructing the clock setting error in the third stage.

In this case, by confirming the modem clock setting or by checking the clock of each modem of the next node group, it is possible to quickly determine which modem has an incorrect clock. Further, if the host computer has just been replaced recently, or if it is known that many computer abnormalities occur in the target network, the diagnosis can be started by instructing the host-related abnormality from the beginning.

In this way, the start node (ROOT)
The time required for the diagnosis can be significantly shortened as compared with the case of performing the diagnosis. Although the above example is a very simple example, it is possible to realize flexible diagnosis in consideration of the experience of maintenance personnel and the characteristics of the network, and it is possible to shorten the maintenance time.

Next, the operation of the inspection unit in the embodiment of the present invention will be described.

In the inspection unit 22, the first process of advancing the diagnosis to the lower node assuming that the instructed node is confirmed,
A second process can be performed in which the designated node is inspected and the diagnosis is advanced based on the result. The distinction can be made by the proper use of the mouse button when designating the node. For example, the right button of the mouse can be used for the first processing, and the left button (or middle button) can be used for the latter processing.

Assuming that the operator gives an instruction for a clock setting error by using a mouse button (right or left), the processing procedure is such that the type of button (right, left,
The coordinates of the display designated as "middle" are sent to the inspection unit 22. The inspection unit 22 determines the node name (node identification name) of the diagnostic tree 11a from the coordinates. A table of node names (node identification names) and coordinates is prepared in advance for this determination. The inspection unit 22 confirms the type of the button, and if the mode is the confirmation mode of the first process, sends the node identification name to the diagnostic node instructing unit 21 to end the execution, and the next instruction is issued from the diagnostic node instructing unit. Wait until it arrives. If the mode requires the inspection of the second processing, the inspection contents of the designated node are confirmed and the inspection is performed.

The inspection unit 22 confirms the inspection name of the target node and executes the inspection program corresponding thereto. For example, in the clock setting error in the diagnostic tree shown in FIG. 2, the inspection name, the inspection result, and the certainty factor thereof are described. For example, when the synchronization check is performed and the result is system synchronization, and the TI line check is performed and the result is the timing source of the line, the certainty factor 0.3 is established. The inspection program sends the obtained results to the inspection department. The inspection unit 22 compares the results obtained from the respective inspection programs with the satisfaction conditions, and gives a certainty factor if they match the satisfaction conditions. All the inspection items for confirming the certainty of the node are confirmed, and the result is sent to the inspection result judging unit 23. That is, the inspection result sent from the inspection unit 22 to the inspection result determination unit 23 is the node name (node identification name) and the certainty factor for each inspection item, and the n of the node.
R1, ~, Ri, ~ Rn (0 <Ri <
The coefficient of 1) is sent. The inspection unit 22 determines whether or not there is another node to be inspected, and returns to the process if there is any and executes the inspection. When the inspection of all the nodes is completed, it waits until the next instruction arrives from the diagnostic node instruction unit 21. The node identification name is sent to the diagnostic node instructing unit 21, the execution is completed, and the process waits until the next instruction arrives from the diagnostic node instructing unit 21.

Table 1 shows an example of inspection items in the embodiment of the present invention.

[0037]

[Table 1] Next, the processing of the inspection result determination unit in the embodiment of the present invention will be described. From the inspection unit 22, as the inspection result, the node name (node identification name) and the node n which is the certainty factor for each inspection item.
R1, ~, Ri, ~, Rn (0 <R
Receive the coefficient of i <1).

Generally, when the start node is instructed, the inspection result received is only for one node. When the diagnosis is advanced from the upper node instead of the start node instruction, the received inspection result is received from a plurality of nodes connected to the lower node of the certain node. The confidence factor Pk of the node name K is calculated by the following calculation.
The certainty factor is given by Pk = 1- (1-Rn). here,
(1-Rn) = (1-R1) × (1-R2) × ... × (1
-Ri) × ... × (1-Rn) is shown. It is determined whether Pk is higher than the reference value. That is, if the reference value is S, if Pk> = Selse
run then. The node considers it unsuitable and displays a message to that effect.
For example, the node is displayed in blue on the display, indicating that the inspection result was bad. Calculate confidence of other nodes. It is determined whether or not there is another node for which the confidence factor calculation is sent from the inspection unit 22, and if there is, the confidence factor is calculated. Select a node whose confidence level exceeds the reference value and has the highest confidence level. The node name (node identification name) is sent to the diagnostic node instructing unit 21.
If the node does not exist, the diagnosis ends, and the fact is displayed on the display.

Upon receiving the node identifier from the inspection result judging unit 23, the diagnostic node instructing unit 21 confirms the lower nodes of the node, that is, four nodes such as the parent modem clock miss in FIG. 2, from the diagnostic tree 11a, and the inspecting unit 22. Ask to perform the inspection. If there is no subordinate node, it means the cause of that node, and the fact is displayed, and the diagnosis ends.

This is repeated.

[0041]

As described above, according to the present invention, the examination can be started from the middle of the diagnostic knowledge base in the state required for the diagnosis, and therefore the unnecessary examination of the hierarchy is eliminated. There is an effect that diagnosis can be performed efficiently.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of the present invention.

FIG. 2 is a diagram showing a configuration of a diagnostic tree of a diagnostic knowledge base in the embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Storage device 2 Central processing unit 3 Input device 4 Display device 11 Diagnostic knowledge base 11a Diagnostic tree 20 Diagnostic device 21 Diagnostic node instruction unit 22 Inspection unit 23 Inspection result determination unit

Claims (2)

[Claims] 1. A knowledge base of diagnostic trees hierarchically arranged so that the nodes are tree-structured, with diagnostic data describing contents for inspecting a fault status corresponding to each failure state as a node. In the method of performing fault diagnosis by sequentially tracing the nodes from upper to lower according to the hierarchical relationship of the diagnosis tree, a node in the middle of the diagnosis tree is designated, and the designated node is used as the starting point. A fault diagnosis method, wherein the fault diagnosis is performed while tracing to a lower level. 2. A knowledge base is a diagnostic tree that is hierarchically arranged so that the nodes have a tree structure, with diagnostic data that describes the content for inspecting the fault status corresponding to each fault status as a node. By comparing the data obtained from the operation of the device under test with the diagnostic data held in this knowledge base according to the hierarchical relationship of the diagnostic tree, the nodes are sequentially traced from upper to lower. In a fault diagnosis device comprising means for selecting a node indicating a probable fault state and means for displaying the selected node as a diagnosis result, means for designating a node in the middle of the diagnosis tree by operation is provided. A fault diagnosis device characterized by the above.