JPH06325016A - Abnormality analyzer - Google Patents

Abstract

PURPOSE:To provide the abnormality analyzer for analyzing/explaining any abnormality caused by an event before the generation of abnormality as well. CONSTITUTION:A trace part 13 traces the relation of cause and effect to the generation of a abnormal parameter by detecting respective parameters as the base of abnormal parameter value decision at reference time based on a normal value and a model and back counting the normal values of these respective parameters. When a parameter detection part 14 with the lapse of time detects any parameter, a normal pattern generation part 15 decides the normal value of the basic parameter at every element time constituting a normal pattern, and a trace derivation part 16 derives the trace by activating the trace part 13 with the basic parameter as the abnormal parameter and with every element time as the reference time.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improvement of an abnormality analysis apparatus for analyzing an abnormality cause of various systems such as a sequence control device and an electric circuit of a plant and explaining the analyzed cause to a user.

[0002]

2. Description of the Related Art Conventionally, an abnormality analysis device for analyzing an abnormality of a sequence control device or an electric circuit of a steel plant has been known. This anomaly analysis device shows the correspondence between the content of the anomaly and its cause.

[Equation 1] Built-in in the form of a so-called If-then rule such as If [content of abnormality] then [cause of abnormality], the content of the input abnormality is collated with this rule, and the cause of abnormality is output. Is.

[0003]

However, since the conventional anomaly analysis apparatus performs anomaly analysis based on the condition judgment based on the If-then rule as described above, it is unexpectedly not prepared as the If-then rule in advance. There was a problem that it was not possible to deal with the anomaly. In addition, since the designer of the abnormality analysis device and the user must obtain the If-then rule based on subjective information such as experience, the performance of the conventional abnormality analysis device is limited to that of the designer or the user. There was a problem that it was limited to the rule construction ability of.

To solve such a problem, a method called model-based inference has been proposed (reference: RMKell).
er ("Applying Knowledge Compilation Thechnique to M
odel-Based Reasoning ", IEEE Expert, Vol.6, No.2, pp.82
-87,1991 / R. Reiter ("A Theory of Diagnosis from Th
e First Principle ", Artificial Inteligence.32,1987,
pp.57-95). In this method, the system to be analyzed is retained in the form of a model that represents the causal relationship of actions among the elements that make up the system, and the causal relationship that leads to anomalous occurrence is analyzed by inference based on this model. , To identify the faulty device.

According to this model-based reasoning, the analysis is performed based on the mechanism (operating principle) of the system, so that the cause of an unexpected abnormality can be investigated. Further, the model used in this method can be constructed by knowledge of the function / structure of the analysis target system and the physical laws, and such knowledge can be easily obtained from system design information and general literature. Therefore, according to this method, the knowledge base used for inference can be easily constructed and the performance of the abnormality analysis device can be easily improved.

The conventional analyzer analyzes the cause of the abnormality, but does not explain the logical structure of the causal relationship. In the present invention, not only the cause of abnormality is analyzed, but also the causal relationship from the cause of abnormality to the occurrence of abnormality is explained. Therefore, the user can determine an appropriate cause of abnormality from the candidates for the cause of abnormality. In addition, the principle of system operation, and the explanation based on inference that returns to the principle, are easy for general users who are not experts to understand, and it is easy for general users to deal with abnormalities.

However, when using model-based reasoning,
Although it is possible to analyze abnormalities in the equipment itself, as seen in sequence control devices, etc., if there is no failure in the equipment itself and an abnormality occurs due to an external factor of the system, it is impossible to analyze the abnormality. There was a point. To solve this problem, a logic search method proposed by Shapiro, Ehud Y. (Reference: "Algrithmic Program De
"Bugging", Thesis (Ph.D.)-Yale University, 1982) is effective. This method finds a contradiction at the end of causality by logically tracing the causal relationship leading to abnormal parameter determination, Since it is possible to search for abnormalities in parameters from outside the system, it is possible to analyze and explain abnormalities caused by external factors.

However, the tracing of the causal relationship by this method is performed at a fixed reference time, and it is possible to analyze and explain the abnormality that is completed within this reference time. However, there are some component devices that make up the system in the past, in which a certain parameter shows a certain change with time in the past as a condition for outputting a normal value. However, it may be due to an event before the occurrence of an abnormality that is usually used as the reference time. In the logic search method, since the search is performed at a certain reference time as described above, it is not possible to analyze and explain the abnormality caused by the event before the reference time.

For example, the abnormality that occurred at time t1 is
If the cause is that the switch-on timing is shifted at a time t2 that is earlier, it is impossible to analyze and explain the cause of the abnormality only by the trace having only the abnormality occurrence time t1 as the reference time.

The present invention has been proposed in order to solve the above-mentioned problems of the prior art, and its purpose is to analyze and explain anomalies caused by events before the occurrence of an anomaly. It is to provide a device.

[0011]

In order to achieve the above object, the abnormality analysis device according to claim 1 comprises a model input means for inputting a model representing a relationship between parameters included in a system to be analyzed, Model storage means for storing the model, actual measurement value input means for inputting the actual measurement value of each parameter for each time, actual measurement value storage means for storing each actual measurement value for each time, and Of these, an abnormal parameter input means for inputting an abnormal parameter that has become abnormal, a reference time input means for inputting a reference time that serves as a reference for the analysis, and a normal value that the abnormal parameter should take at the reference time. Based on the normal value input means for inputting, and the normal value and the model, each of the basis for determining the abnormal parameter value at the reference time. By detecting the parameter, by back-calculating the normal value of each of these parameters, tracing means for tracing the causal relationship leading to the occurrence of the abnormal parameter, and of the basic parameters that became the basis of the parameter value determination of each of the parameters A temporal parameter detecting means for detecting a temporal parameter that takes a normal value when a value for each time changes in a predetermined temporal normal pattern, and a normal value for the basic parameter at each element time constituting the normal pattern A normal pattern generating means for determining, a trace deriving means for deriving a trace by activating the tracing means with the basic parameter as the abnormal parameter and each element time as the reference time, and the cause and effect of the abnormal parameter determination. Of the measured value of the parameter located at the beginning of the relationship Then, a cause candidate detecting means for detecting as the cause candidate of the abnormality a value that does not match the back-calculated normal value of the parameter, and an abnormality explanation generation for generating an abnormality explanation related to the cause of the abnormality based on at least the cause candidate. And a means for outputting the abnormality description.

According to a second aspect of the present invention, in the abnormality analysis apparatus according to the first aspect, a match pattern detecting means for detecting a match pattern matching the normal pattern from the measured value of the basic parameter, and the match pattern. A second trace derivation means for deriving a trace by activating the trace means with a time that is earlier than the reference time as the reference time, which is earlier than the normal pattern, as the reference time. .

According to a third aspect of the present invention, in the abnormality analyzing device according to the first aspect, drop detection means for detecting a drop in the measured value of the abnormal parameter from the normal value and a time of this drop, and Third trace derivation means for deriving a trace by activating the trace means with the drop time as the reference time.

Further, the invention according to claim 4 is the abnormality analyzing device according to claim 1, wherein each of the parameters to be the basis for determining the parameter for which the actually measured value matches the back-calculated normal value is the basis for determining the parameter. It is characterized in that a suppression means for suppressing the trace is provided.

According to a fifth aspect of the present invention, the abnormality analysis device according to the fourth aspect is characterized in that a suppression selecting means for selecting whether or not to perform the suppression is provided.

[0016]

The present invention having the above structure has the following functions. That is, in the invention of claim 1, when the temporal parameter is detected in the trace, the normal value of the basic parameter at each element time forming the normal pattern is determined as the basis of the temporal parameter determination. Then, since the trace is derived for each element time, it is possible to find the starting point of the cause and effect that the measured value deviates from the normal value at each element time, that is, the cause of the abnormality.

As described above, according to the first aspect of the present invention, even with respect to a temporal parameter affected by a temporal change of the basic parameter in the past, tracing is performed retroactively to analyze the cause of the abnormality. It is also possible to analyze and explain anomalies caused by events before the occurrence of an anomaly.

Further, in the invention of claim 2, since the time difference in which the coincident pattern is advanced from the normal pattern and the trace in which the reference time is advanced are derived, the cause of the earlier occurrence of the normal pattern is analyzed. can do.

Further, according to the third aspect of the invention, since the trace is performed with the time when the measured value of the abnormal parameter falls from the normal value as the reference time, the cause of the past drop of the abnormal parameter from the normal value can be analyzed. it can.

Further, in the invention of claim 4, since the trace target range is limited to the range in which the measured value is abnormal,
Anomaly analysis can be made efficient.

Further, in the invention of claim 5, since it is possible to select whether or not to inhibit, there is a possibility that a parameter once blocked needs to be retraced by a trace derived thereafter due to the nature of the analysis target system. If so, the action of the deterrent can be stopped.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An abnormality analysis apparatus (hereinafter referred to as "this apparatus") which is an embodiment of the present invention will be specifically described below with reference to the drawings. It should be noted that this device is realized on a computer, and each function of this device is realized by operating the computer in a predetermined procedure expressed in the form of a program. Therefore, the present apparatus will be described below assuming a virtual circuit block having each function of the apparatus.

(1) Configuration of the Embodiment ... FIG. 1 First, FIG. 1 is a block diagram showing the configuration of the present apparatus.
That is, the present device, as shown in this figure, includes a keyboard 1 and a communication line 2 for input, a display device 3 for output,
And a processing unit 4 for processing information. Although not shown, the communication line 2 is connected to a sensor provided in each part of the system to be analyzed, and the actual measurement value of each parameter included in the system at each time is the communication line 2. Is input to this apparatus via.

The processing unit 4 also includes an I / O control circuit 5 and
A model input unit 6 that takes in data that constitutes a model representing a relationship between parameters included in the analysis target system from the keyboard 1 and a model storage unit 7 (corresponding to the model storage unit) that stores the model. Have Further, the present apparatus is provided with an actual measurement value input unit 8 for taking in the actual measurement value of each parameter for each time from the communication line 2.
And a measured value storage unit 9 that stores the measured values at each time.
(Corresponding to the measured value storage means). In addition, the present apparatus has an abnormal parameter (which parameter is abnormal) among the above parameters.
Parameter input unit 10 for fetching a reference time from the keyboard 1, a reference time input unit 11 for fetching a reference time to be a reference for analysis from the keyboard 1, and a normal value for the abnormal parameter at the reference time to the keyboard 1 It has a normal value input unit 12 for taking in from.

The model input unit 6, the abnormal parameter input unit 10, the reference time input unit 11, and the normal value input unit 12 are
With the keyboard 1 and the I / O control circuit 5, respectively,
The model input means, the abnormal parameter input means, the reference time input means, and the normal value input means are configured. Also,
The actual measurement value input unit 8 constitutes the actual measurement value input means together with the communication line 2 and the I / O control circuit 5.

Further, the present apparatus detects, on the basis of the normal value and the model, each parameter which is the basis for determining the abnormal parameter value at the reference time, and calculates the normal value of each of these parameters backwards. A trace unit 13 (corresponding to the trace means) that traces the causal relationship leading to the generation of the abnormal parameter.

In addition, the present apparatus is a temporal parameter that takes a normal value when the value of the basic parameter, which is the basis of the parameter value determination among the parameters, changes with a predetermined normal pattern over time. Parameter detector 14 (corresponding to the parameter detector) and a normal pattern generator 15 (normal) to determine the normal value of the basic parameter at each element time of the normal pattern. Pattern generating means), the basic parameter as the abnormal parameter, and each element time as the reference time, and the trace deriving unit 16 that activates the trace unit 13 to derive a trace (in the trace deriving unit). Equivalent)).

Further, the present apparatus includes a matching pattern detecting section 17 (corresponding to the matching pattern detecting means) for detecting a matching pattern matching the normal pattern from the measured value of the basic parameter, and the matching pattern detecting section 17 A trace derivation unit 18 that activates the trace unit 13 and derives a trace by using a time that is earlier than the difference of the preceding time with respect to the normal pattern and the reference time as the reference time (the second derivation).
Equivalent to the trace derivation means).

Further, the present apparatus detects a dropout of the measured value of the abnormal parameter from the normal value and a dropout detection section 19 (corresponding to the dropout detection means) for detecting the time of the dropout, and the dropout time. Trace derivation unit 2 that activates the trace unit 13 as a reference time to derive a trace
It has 0 and.

Further, the present apparatus has a suppressing section 21 (corresponding to the above-mentioned suppressing means) for suppressing the tracing from the parameter whose actual measured value matches the back-calculated normal value.

Further, the present apparatus detects, as a candidate for the cause of the abnormality, among the parameters located at the starting point of the causal relationship for determining the abnormal parameter, the measured value of the parameter does not match the back-calculated normal value. Cause candidate detection unit 2
2 (corresponding to the cause candidate detecting means) and a cause candidate storage unit 23 for storing these cause candidates. The apparatus also includes a trace process storage unit 24 that stores all the processes of the trace.

The apparatus further includes an abnormality explanation generation unit 25 (corresponding to the abnormality explanation generation means) for generating an abnormality explanation relating to the cause of the abnormality based on the entire process of the cause candidate and the trace, and the abnormality. The display control unit 26 for displaying the explanation on the display device 3. The display device 3 and the display control unit 26 constitute the output means.

(2) Operation of the embodiment ... FIGS. 2 and 3 The abnormality analysis by the present apparatus having the above-mentioned configuration is performed as follows. Here, FIG. 2 is a flowchart schematically showing the entire abnormality analysis procedure in the present apparatus.

In this procedure, first, necessary data is input (step 21). That is, in this data input, the user inputs the model through the model input unit 6, and the model storage unit 7 stores this model. Further, the measured value input unit 8 receives the measured value of each parameter of the analysis target system from the communication line 2 at each time, and the measured value storage unit 9 stores each measured value at each time. In addition, the user inputs the abnormal parameter input unit 10 and the reference time input unit 1
1. The abnormal parameter, the reference time, and the normal value of the abnormal parameter at the reference time are input via the normal value input unit 12.

Subsequently, the dropout detection unit 19 detects the dropout from the normal value in the measured value of the abnormal parameter and the time of this dropout (step 22). A trace is derived by activating the trace unit 13 with the drop time as a reference time and the abnormal parameter initially input by the user as a starting point (step 23). With this trace, it is possible to analyze the cause of the actual measurement value changing (dropping) from the normal value to the abnormal value. No dropout was detected in step 22 or this derived trace (step 2
When step 3) is completed, the trace unit 13 performs the original trace (step 24).

Here, the trace performed in step 24 detects the normal values initially input by the user and the respective parameters that are the basis of the abnormal parameter value determination at the reference time, based on the model, and the respective parameters are detected. Is a process of tracing back the causal relationship leading to the occurrence of the abnormal parameter by back-calculating the normal value of. Then, the trace in step 23 is obtained by shifting only the reference time of the trace in step 24 to the drop time.
The trace in this step 23 can be performed after step 24, but is performed in connection with step 24 in that it is based on the original trace specified by the user (step 24). It is something. Therefore, in the present embodiment, for convenience, step 2
Treated as a "derivative" trace from the 4 trace.

That is, in the present specification, “derivation” means executing another trace with respect to a certain trace with respect to a reference time or a starting point (abnormal parameter) different from that trace. When a trace is derived, the reference time, start point, current parameter, work area in memory, etc. related to the pre-derivation trace are saved in a specified memory area, and when the derived trace ends, the pre-derivation trace Will continue.

In the present embodiment, such trace derivation is realized by using so-called recursive call. Here, the recursive call means that a certain procedure re-executes the procedure itself, and when the procedure is realized by a computer program, it means that a certain routine or function calls the routine or the function itself. means. FIG. 3 is a flowchart showing a specific procedure of the trace shown in steps 23 and 24 of FIG. 2, and is based on such recursive call.

It should be noted that the plurality of traces in this embodiment may be executed in parallel at the same time. Such simultaneous parallel execution uses a time division system (multitask system) or a plurality of CPUs. It can be realized by the parallel processing system.

Further, by recursive call, trace derivation can be performed in multiples (nesting). For example,
It is also free to further derive a third trace based on this second trace during a second trace derived based on one first trace.

By the way, the present apparatus is premised on that, in the analysis target system, one parameter may have a plurality of parameters on the upstream side of its causal relationship.
Since such an upstream-downstream relationship can be established in multiple stages, the possibility of the parameter back calculation in the trace of the present embodiment starts from the abnormal parameter serving as the root,
A search space having a tree structure is constructed. The end of each branch of this tree structure is the starting point of the causal relationship for determining an abnormal parameter. In order to trace the search space without omission, the trace unit 13 forms a map representing the search space on the memory and follows the map with a constant pointer. The parameter currently pointed to by this pointer will be referred to as a current parameter hereinafter.

The data corresponding to each parameter on the map includes at least 1-bit flag data. The flag data is used to indicate a parameter that has been processed and does not need to be searched again. The suppression unit 21 mainly sets the parameter whose actual measurement value matches the back-calculated normal value. This flag has the effect of blocking traces ahead of the flagged parameter, since the flagged parameter is ignored in the search for branch destinations in the trace. Therefore, the operation of setting this flag will be referred to as "block" hereinafter.

The procedure shown in FIG. 3 constitutes a trace procedure by being repeatedly executed. That is, in the procedure shown in FIG. 3, first, the trace unit 13 is not blocked among the parameters that are the basis for determining the current parameter based on the model, that is,
One branchable one is determined (step 31), the pointer is advanced to this parameter (step 35), and the normal value of the new current parameter is back-calculated so as to generate the normal value of the immediately preceding current parameter (step 35). Step 43). It should be noted that this back calculation can be realized in a model having a simple parameter value by simply reading out predetermined data from the model.

Then, the cause candidate detection unit 22 judges whether or not the measured value of the current parameter matches the normal value of the current parameter (step 47). If they do not match, the current parameter is the starting point of the causal relationship for determining the abnormal parameter. It is determined whether or not (hereinafter referred to as "causal starting point") (step 48). In the case of the causal starting point, since the current parameter is the cause of the abnormality, the cause candidate storage unit 23 stores the current parameter as the cause candidate of the abnormality (step 4).
9). Since it is not necessary to retrace the current parameters registered as cause candidates in this way,
The suppression unit 21 blocks the current parameter (step 41). Further, in this case, since the causal starting point has already been examined and another branch destination search from the immediately preceding parameter is necessary, the tracing unit 13 retreats the pointer to the immediately preceding parameter (step 42). , A new branch destination is searched (step 31).

Prior to the comparison between the normal value and the actually measured value, the trace unit 13 determines whether or not the current parameter is a self-hold parameter that changes based on the parameter already stored as a cause candidate (step 4).
4). Such determination of self-hold can be made by determining whether or not the connection relationship between the parameters represented in the model satisfies a predetermined condition for establishing self-hold. In the case of self-holding, the current parameter is meaningless as a cause candidate because the cause of the abnormality is already known as a cause candidate. Therefore, the inhibiting unit 21 blocks the current parameter (step 45). In this case, the current parameter has already been examined, and another branch destination search from the immediately preceding parameter is necessary. Therefore, the pointer is moved back to the immediately preceding parameter (step 46) and a new branch destination is searched. (Step 31).

In step 47, even when the measured value matches the normal value, there is no cause of abnormality ahead of the current parameter, so the inhibiting unit 21 blocks the current parameter (step 45). In this case as well, the current parameter has already been examined, and another branch destination search from the immediately preceding parameter is necessary.
Moves the pointer backward (step 46) and searches the branch destination again (step 31).

If the actual measurement value of the current parameter is not a normal value (step 47) and it is not the causal starting point (step 48), since the trace before the current parameter is necessary, the procedure returns to step 31. The trace unit 13 searches for a branch destination from the current parameter. In the branch destination detection in step 31, the blocked parameter has already been considered and is ignored as the branch destination.

When all branch destinations from a certain current parameter are blocked (step 31), it is meaningless to consider the current parameter itself.
The suppression unit 21 also blocks the current parameter itself (step 32). Then, the trace unit 13
It is judged whether or not the current parameter is the trace start point (step 33), and if it is the trace start point, it is considered that the trace is completed and the control is returned to the calling side. On the other hand, if it is not the start point, it is necessary to search for a branch destination from a parameter closer to the start point than the current parameter.
Moves the pointer backward (step 34) and searches for a new branch destination (step 31).

After advancing the pointer, the temporal parameter detection unit 14 determines whether or not the current parameter is a temporal parameter (step 36). If the current parameter is a temporal parameter, the normal pattern generation unit 15 determines the normal pattern as described above. The normal value of the basic parameter at each constituent time is determined (step 37). Then, the trace deriving unit 16 derives a trace by activating the trace unit 13 using the basic parameter as the abnormal parameter and each element time as the reference time (step 38). As a result, it is possible to find the starting point of the cause of the measured value deviating from the normal value at each element time, that is, the cause of the abnormality.

Then, the matching pattern detecting section 17 judges whether or not a matching pattern matching the normal pattern is detected from the measured values of the basic parameters (step 3).
9) If a matching pattern is detected, the trace derivation unit 18 activates the tracing unit 13 by using the preceding time difference of the matching pattern with respect to the normal pattern and a time that is past the reference time as the reference time. By deriving a trace (step 40). As a result, it is possible to find out the cause of the occurrence of the normal pattern.

In the above trace, the trace process storage unit 24 stores the trace process for all the parameters to be traced. After the end of the trace, the abnormality explanation generation unit 25 logically configures and generates an abnormality explanation related to the cause of the abnormality based on the cause candidate and the tracing process to reach the cause candidate (step 2).
5). In addition, in the logical configuration of the abnormality explanation, the abnormality cause is expressed by ANDing or ORing each element device and its parameter leading to the discovery of the cause candidate. The abnormality description generated in this way is displayed on the display device via the display control unit 26 (step 26).

(3) Actual Example ... FIGS. 4 to 7 Next, as an actual example of the abnormality analysis by this apparatus, the abnormality analysis of the sequence control apparatus will be described. FIG. 4 is a ladder diagram showing a portion to be analyzed in the sequence control device. In this ladder diagram, the contacts Sw1 to Sw4 and p1, which are the component devices that form the sequence control device, are connected by a signal line to the coils C1 and C2 and the connection point OR. In this diagram, the contacts correspond to switches. However, the coil corresponds to the excitation valve.

In each of these component devices, the flow of signals and the processing are defined as a causal relationship, whereby the sequence control in the sequence control device is realized. That is, in the sequence control device of FIG. 4, the power (signal) flows from left to right on the horizontal line, and the power flow is opened / closed depending on the open / close state of each contact (logical product AND
Is done). Here, in FIG. 4, each element device and its parameter are surrounded by a circle, and this circle is called a node. Each node is connected to other nodes by parameters: input, output and state parameters. Further, the range of each of these parameters is a binary value of 1 and 0, and expresses on and off of the current, respectively.

In FIG. 4, the contacts Sw1 to Sw4 are
It has a function of generating 1 as the value of the output parameter out only when the values of both the input parameter in and the state parameter st are 1. Also, each coil C1, C2
Has a function of generating 1 as the value of the state parameter st when the value of the input parameter in is 1. Further, the contact p1 is a special contact called a positive direction change detection contact, and the value 1 of the output parameter out is generated for one control cycle time only when the value of the input parameter in changes from 0 to 1. Have a function.

The connection point OR performs a logical sum, and if either of the values of the two input parameters in1 and in2 is 1, the value of the output parameter out is 1. At the contact point Sw3, the value of the state parameter st is equal to the value of the state parameter st of the coil C1.
A self-hold circuit is provided from 3 to the coil C1. This self-hold circuit keeps the state once the coil C1 is turned on and turned on unless the contact Sw2 is turned off and turned off.

Next, Tables 1 to 3 are examples of description of causal relationships between the respective elements of the sequence control device of FIG. 4, and a set of these causal relationships is a model. This model expresses the relationship between parameters with each element device as a unit. That is, the causal relationship of the contact point Sw2 has the content shown in Table 1 below.

[0057]

[Table 1]

The causal relationship of the coil C1 is shown in Table 2 below.
The contents are shown in.

[0059]

[Table 2]

The causal relationship of the positive direction change detection contact p1 is as shown in Table 3 below.

[0061]

[Table 3] In Tables 1 to 3 above, "name" is the name of the element device, "parameter" is a list of parameters belonging to the device, and "connection" is the name of another device to which the parameters are connected. Further, a list of connection target parameters in the other device, “function”, represents the function of the device by a relational expression between parameters of the device.

[Example 1] FIG. 5 is a time chart showing measured values of the respective parameters before and after time t4 when an abnormality (Example 1) occurs in the sequence control device. According to the notation rule of the time chart of FIG. 5, when the line is drawn from time t to t ′ (t <t ′), the range indicated by the line is the time t on the past side.
However, the time t ′ on the future side is not included. A dotted line having the same width as the line showing the value of each parameter is a normal value at each time.

The abnormality of Example 1 is that the value of the state parameter st of the coil C1 should be the normal value 1 at time t4.
Is 0. In Example 1, the time t4 is used as the reference time and the coil C1 is used as the abnormal parameter.
1 is input as the normal value of the abnormal parameter at the reference time.

First, since the measured value of the abnormal parameter does not detect the drop from the normal value, the pointer advances to the contact Sw2 on the cause side, and the state parameter st of the coil C1 is changed.
Normal value of 1 (that is, output parameter ou of contact Sw2)
The normal value 1 of the state parameter st of the contact Sw2 and the normal value 1 of the input parameter in of the contact Sw2 are calculated back based on the normal value of t) and the model. Here, contact point Sw2
Since the measured value 1 of the state parameter st is equal to the normal value 1 of the same parameter, the same parameter is blocked, and the coil C2 and contact Sw4 that are continuous to the cause side of the same parameter are connected.
Is not traced. On the other hand, since the measured value 0 of the input parameter in of the contact point Sw2 is different from the normal value 1 of the same parameter, the trace proceeds in the same parameter direction.

That is, the pointer advances to the connection point OR continuous to the cause side of the input parameter in of the contact Sw2,
Based on the normal value 1 of the input parameter in of the contact point Sw2 (that is, the normal value of the output parameter out of the connection point OR) and the model, the input parameter in of the connection point OR
Each normal value 1,1 of 1, in2 is calculated back. Here, the actual measurement value 0 of the input parameter in1 of the connection point OR is different from the normal value 1 of the same parameter, and the actual measurement value 0 of the input parameter in2 of the connection point OR is different from the normal value 1 of the same parameter. It is performed sequentially in both directions.

That is, the pointer first advances from the connection point OR to the contact point p1 continuous to the cause side of the input parameter in1 of the connection point OR, but since the contact point p1 is an element including a temporal parameter, based on the model. , A normal pattern for the output parameter out of the contact p1 (that is, the input parameter in1 of the connection point OR) to be 1 at the time t4 is a pattern in which the input parameter in of the contact p1 changes from 0 to 1 at the time t4. Is generated (indicated by a dashed circle in FIG. 5). This pattern is
In other words, the input parameter in of the contact p1 (that is, the output parameter out of the switch Sw1) is the time t3.
Means 0 and 1 at time t4. Therefore, a trace having a reference time of time t3 and a normal value of 0 is recursively derived from the contact point p1 as a starting point.

In the trace whose reference time is the time t3,
The pointer advances to the contact point Sw1 continuing to the cause side of the input parameter in of the contact point p1, and the contact point Sw1 at time t3.
Of the contact Sw1 at the time t3 based on the normal value 0 of the output parameter out of
The normal value of 0 is calculated backward. The normal value of the input parameter in of the contact point Sw1 is always defined as 1 by the model. Here, the actual measurement value 1 of the state parameter st of the contact point Sw1 at the time t3 is different from the normal value 0 of the same parameter at the same time point, and the contact point Sw1 is one of the causal starting points in the sequence control device.

Therefore, the information that "the measured value 1 of the state parameter st of the contact point Sw1 at time t3 is different from the normal value 0 of the same parameter at the same time" is registered as a cause candidate, and the derived trace ends. To do.
From the description based on this cause candidate, the user indicates that the state parameter st of the switch Sw1 is 0 to 1 at time t4, for example.
Although it had to be set to 1, it has already changed to 1 at t3, and it can be determined that this change is too early to cause the abnormality.

Subsequently, time t4 which is the reference time before derivation
Is the input parameter in of contact p1
(That is, the normal value of the output parameter out of the switch Sw1) is set to 1 and the process is continued. That is, the pointer is the contact point S that continues to the cause of the input parameter in of the contact point p1.
The process proceeds to w1 and, based on the normal value 1 of the output parameter out of the contact Sw1 at time t4 and the model, the contact Sw
The normal value 1 of the input parameter in 1 is calculated back. Here, the actually measured value 1 of the input parameter in of the contact point Sw1 at time t4 is the normal value 1 of the same parameter at the same time.
Therefore, the same parameter is blocked, the pointer returns to the coil C1 which is the starting point of the trace, and the trace ends.

Further, the same matching pattern as the normal pattern from 0 to 1 is detected from the measured value of the input parameter in of the contact p1 (shown by the solid circle in FIG. 5), and the time is t1. Therefore, a trace is derived with the coil C1 as the starting point and the reference time as the time t1.

In this trace, the actually measured value of the input parameter in of the contact Sw2 coincides with the normal value.
2 the input parameter in is blocked, while the contact S
The measured value of the state parameter st of w2 is different from the normal value.
Therefore, the same trace as above is continued for the branch destination of the state parameter st of the contact Sw2, resulting in time t1.
The measured value of the state parameter st of the contact Sw4 at 1 is 1
The point that is not is stored as a cause candidate.

From the connection point OR, the pointer advances to the contact point Sw3 continuous to the cause side of the input parameter in1 of the connection point OR. Here, since the state parameter st of the contact Sw3 is equal to the state parameter st of C1 (self-hold circuit), the actually measured value of the input parameter in of the contact Sw3 is 1, which is equal to the normal value, and the state parameter st of the contact Sw3 is Since it is the state parameter st itself of C1, the value of the state parameter of the contact point Sw3 is naturally 0 with respect to the normal value 1. Therefore, no cause of abnormality is recognized in this direction, and the input parameter in of the contact Sw3 is blocked. As a result, even if the pointer returns to the connection point OR, the contact Sw2, and the coil C1, the branch destination that is not blocked cannot be found, and the trace ends.

FIG. 6 is a tree representation of the above tracing process in the first example. In this example, as an explanation of abnormality, 1) the state parameter st of Sw1 is 1 at time t3, and 2) Sw is set at time t1.
It is output that the state parameter st of 4 is not 1, and from this explanation of the abnormality, the cause of the abnormality is that the user can 1) quickly set the state parameter of Sw1 to 1.
2) It is possible to determine that it is the cause of abnormality that it is slow to set the state parameter of Sw4 to 1.

As described above, in this embodiment, since a plurality of explanations of abnormality are output for one abnormality, the designer / adjuster selects the abnormality explanation considered to be the most rational and copes with the abnormality. be able to.

[Example 2] FIG. 6 is a time chart showing measured values of the respective parameters before and after time t7 when an abnormality (Example 2) occurs in the sequence control device. In this example, the value of the state parameter st of the coil C1 which should be the normal value 1 at time t7 is 0.

In Example 2, the time t7 is set as the reference time,
As the abnormal parameter, the state parameter st of the coil C1
1 as the normal value of the abnormal parameter at the reference time
Enter. Example 2 is different from Example 1 in that a drop from the normal value is detected in the coil C1, but
Others can be analyzed by a processing procedure that is generally the same as that of the first example. Therefore, first, the detection of the drop from the normal value in the coil C1 and the subsequent derivative trace will be described.

That is, the drop from the normal value is the change from the normal value 1 to other values, specifically, the abnormal value 0,
This change is found at time t4. Therefore, when the reference time is set to the time t4 and the start point is set to the coil C1 and the trace is derived and executed, as a result, as a cause candidate,
The state parameter st of the contact point Sw4 is 0 at time t4.
Is stored. Based on the abnormality explanation based on this cause candidate, the user, for example, sw4 at time t4.
It is possible to determine that the self-hold is released because the value of the state parameter st of 0 has temporarily become 0, and the value of the state parameter st of the coil C1 has become 0.

On the other hand, in the trace having the time t7 determined by the user as the reference time, the point where the state parameter of Sw1 is 1 at time t6 is stored as the cause candidate.

(4) Effects of the Embodiment As described above, in the present embodiment, tracing is performed retroactively for each element time relating to the temporal parameter (steps 37 and 38), so that before the occurrence of an abnormality. It is possible to analyze and explain anomalies caused by the above events. In addition, in this embodiment, with respect to the temporal parameters,
Since the time difference in which the matching pattern is advanced from the normal pattern and the trace in which the reference time is advanced are derived (steps 39 and 40), it is possible to analyze the cause of the occurrence of the normal pattern in the past.

Further, in the present embodiment, since the trace is performed with the time when the measured value of the abnormal parameter falls from the normal value as the reference time (steps 22 and 23), the cause of the past drop of the abnormal parameter from the normal value is determined. Can be analyzed. Further, in the present embodiment, the target range of tracing is limited to the range in which the actual measurement value is abnormal (step 41), so that the abnormality analysis can be made efficient.

In particular, in the present embodiment, a plurality of cause candidates are explained, so that the user can select the explanation that is the most rational among them.

Further, although the conventional abnormality analysis device did not output the abnormality explanation including the logical configuration, in the present embodiment,
Since the logical configuration for explaining the abnormality is performed by AND connection or OR connection, it becomes easier for the user to determine the cause of the abnormality.

(5) Other Embodiments The present invention is not limited to the above embodiments,
Since the embodiment can be freely changed, the following other examples are also included. For example, the model may be configured in units of system parameters, or may be configured in units of element devices. The measured values of each parameter may be input from the keyboard instead of the communication line. Further, a plurality of abnormal parameters may be designated at the same time. Further, the specific numerical value of each parameter is not limited to 1 or 0. Further, the normal pattern and the matching pattern are not limited to one change in numerical value between two values. Further, in the derivation of the trace related to the temporal parameter, the reference time of each trace does not need to be strictly the element time, but may be a time having a certain relationship with the element time, such as immediately before or immediately after that.

Further, with respect to the drop of the abnormal parameter from the normal value, it is assumed that the value is not the normal value at the time before the occurrence of the abnormality, the change pattern in that case is calculated, and the measured value is changed to this change pattern. It can also be done by searching for a matching pattern.

Further, the abnormality analysis apparatus of the present invention may be provided with a deterrent selecting means for selecting whether or not the deterrence is to be performed. In this way, the action of the deterrent means can be stopped when there is a possibility that a trace that is once blocked may require another trace due to the nature of the analysis target system. That is, in general, when a measured value and a normal value match for a certain parameter, it can be determined that the parameter is not abnormal, and it is not necessary to trace that parameter first. However, when a certain parameter is blocked at a certain time, it may be necessary to trace the parameter again at that time due to a change in the reference time related to a temporal element. In such a case, if the inhibition selecting means enables tracing again, it becomes possible to operate the abnormality analysis device according to the property of the analysis target system.

Further, although the abnormality analyzing apparatus of the above-mentioned embodiment is realized on a computer, all or part of the function may be realized on a dedicated electronic circuit.

[0087]

As described above, according to the present invention, since it is possible to provide the abnormality analysis device that analyzes and explains the abnormality caused by the event before the occurrence of the abnormality, the abnormality analysis becomes easy. .

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an abnormality analysis device according to an embodiment of the present invention.

FIG. 2 is a flowchart schematically showing an abnormality analysis procedure in the same apparatus.

FIG. 3 is a flowchart showing a specific tracing procedure in the apparatus.

FIG. 4 is a ladder diagram showing a configuration of a sequence control device in an example of an embodiment of the present invention.

FIG. 5 is a chart diagram (actual example 1) showing measured values of respective parameters of the sequence control device when an abnormality occurs in the same example.

FIG. 6 is a chart diagram (actual example 2) showing measured values of respective parameters of the sequence control device when an abnormality occurs in the same example.

FIG. 7 is a diagram showing a tracing process in Example 1 of the embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... Keyboard 2 ... Communication line 3 ... Display device 4 ... Processing part 5 ... I / O control circuit 6 ... Model input part 7 ... Model storage part 8 ... Measured value input part 9 ... Measured value storage part 10 ... Abnormal parameter input part 11 ... Standard time input section 12 ... Normal value input section 13 ... Trace section 14 ... Temporal parameter detection section 15 ... Normal pattern generation section 16, 18, 20 ... Trace derivation section 17 ... Matching pattern detection section 19 ... Drop detection section 21 ... Suppression unit 22 ... Cause candidate detection unit 23 ... Cause candidate storage unit 24 ... Trace process storage unit 25 ... Abnormality explanation generation unit 26 ... Display control unit S ... Each step of the procedure

Claims (5)

[Claims] 1. A model input unit for inputting a model representing a relationship between parameters included in a system to be analyzed, a model storage unit for storing the model, and an actual measurement value of each parameter for each time. An actual value input means for inputting; an actual value storage means for storing each actual value at each time; an abnormal parameter input means for inputting an abnormal parameter that has become abnormal among the respective parameters; Reference time input means for inputting a reference time to be a reference of, a normal value input means for inputting a normal value that the abnormal parameter should take at the reference time, based on the normal value and the model, Detecting each parameter that was the basis for determining the abnormal parameter value at the reference time, and calculating the normal value of each of these parameters backward. Therefore, the tracing means for tracing the causal relationship leading to the occurrence of the abnormal parameter, and the value for each time of the basic parameter, which is the basis for determining the parameter value among the parameters, changed in a predetermined normal pattern over time. A temporal parameter detecting unit that detects a temporal parameter that normally takes a normal value, a normal pattern generating unit that determines a normal value of the basic parameter at each element time forming the normal pattern, and the basic parameter that is abnormal Among the actual measurement values of the parameters located at the starting point of the causal relationship of the abnormal parameter determination, and a trace derivation means for deriving a trace by activating the tracing means, with each of the element times as the reference time as the reference time, Those that do not match the above-mentioned normal value calculated backward A cause candidate detecting means for detecting as a cause candidate of the abnormality; an abnormality explanation generating means for generating an abnormality explanation concerning the cause of the abnormality based on at least the cause candidate; and an output means for outputting the abnormality explanation. Anomaly analysis device. 2. A match pattern detection unit that detects a match pattern that matches the normal pattern from the measured value of the basic parameter, and a time that is earlier than the reference time by which the difference in preceding time of the match pattern from the normal pattern is found. 2. The abnormality analysis device according to claim 1, further comprising: a second trace derivation unit that derives a trace by activating the tracing unit with the reference time as. 3. A drop detection means for detecting a drop in the measured value of the abnormal parameter from the normal value and the time of this drop, and tracing by activating the trace means with the drop time as the reference time. 3. An abnormality analysis apparatus according to claim 1, further comprising a third trace derivation means for deriving. 4. The suppressing means for suppressing the tracing to each of the parameters, which is the basis of the determination of the parameter in which the actual measurement value matches the back-calculated normal value, is provided. The described abnormality analysis device. 5. The abnormality analysis apparatus according to claim 4, further comprising a deterrent selecting means for selecting whether or not the deterrence is performed.