JPS63313208A - Plant diagnostic method - Google Patents

Abstract

PURPOSE:To eliminate a retrieval miss, and to efficiently execute a diagnosis by storing in advance the design information of a control system safety system of a plant, as a data in a computer, and utilizing it for the diagnosis. CONSTITUTION:A diagnostic method of a plant is provided with frame type data bases 1-3 which have stored design data, a frame retrieving device 4 for retrieving said data bases, and a demonstration procedures executing device 5. In this state, a flow of its diagnosis inputs 6 an observation data, from which a signal for verifying whether a discrepancy exists or not is set 7. Also, an input which requires the verification is estimated by using a frame data. Moreover, while referring to a predicative frame data by using the observation data, it is compared with a measured value and whether it is normal or abnormal is decided 10, and also, its result is displayed 11. In such a way, by preparing in advance a procedure of an input calculation in accordance with a form of the observation data, it is possible to cope with various events.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention relates to a plant diagnosis method for diagnosing the presence or absence of abnormalities occurring in control system electronic equipment of large-scale power generation plans 1 to 1. be.

(Conventional technology) Regarding the abnormality diagnosis system for large-scale power generation plants,
Many proposals have been made so far.

Next, an example of a typical plant diagnosis system will be explained.

(1) A plant diagnosis system that identifies abnormalities by importing observation time series data from major observation points online, converting it into a frequency domain pattern using noise analysis methods, and comparing it with normal patterns. is proposed.

In this case, if the pattern of the abnormality is known, the cause of the abnormality can be identified by comparison with that pattern.

(2) Using the so-called model comparison method, which involves preparing a model of plant equipment, inputting observed time-series data into the input of the model, predicting the output, and detecting abnormalities by comparing this with actual measured values. A plant diagnostic system has been proposed. In this case, there are two types of models: one in which a physical model is prepared in advance, and one in which it is automatically created from observation data using an AR model or the like.

(3) A plant diagnosis system has been proposed that summarizes the causal relationships of each device in a plant as a cause-and-effect tree (so-called 0CT) and traces this OCT from observed events to identify causes or predict results. Zarani, OCT
A plant diagnosis system applying knowledge engineering has also been proposed, which compiles the information into so-called production-type rules and creates and diagnoses a chain of events equivalent to CCT according to the event.

As described above, conventional plant diagnosis systems are suitable or unsuitable depending on the type of abnormal event that occurs, and are usually systems that are based on online real-time operation.

(Problems to be Solved by the Invention) One of the problems with such conventional diagnostic systems is that they deal with real-time issues. Because high speed is required, plant models and OCT used for diagnosis are either fixed or have simplified the actual equipment configuration to some extent. Therefore, it is often difficult to obtain the detailed information necessary to investigate the cause of an actual problem. Furthermore, when solving such practical problems, it is often difficult to apply a fixed diagnostic model due to missing or misunderstood observation information.

The present invention has been made in view of the above circumstances, and its purpose is to create a detailed and flexible configuration of various plant databases, which are the basis of diagnosis, by applying concepts used in knowledge engineering, and to create a diagnostic system. The objective is to provide a plant diagnosis method that can be used as a basis for

[Structure of the Invention] (Means and Actions for Solving the Problems) In order to achieve the above object, the present invention provides design reports for the electronic equipment for the control system and safety system of power generation plan 1 to , instrument frame data representing information such as setting parameter values, line frame data representing instrument connection relationships, and instruction manual frame data representing instrument functions. Then, from among the plurality of observed values obtained from the control system and safety system of the power plant, the input observed value necessary for predicting an arbitrary observed value is searched from the frame type database, and the input observed value is searched for. The device is characterized in that the output of the device is predicted from the observed value through input/output calculation processing of a frame type database, and the predicted value is compared with the observed value to determine whether or not there is an abnormality in the device. It is something.

As described above, the present invention stores design information of a plant's control system safety system as data in a computer, and utilizes this for diagnosis. In particular, not only the parameter values and connection relationships of the instruments used as design information, but also procedural data related to their functions (input/output relationship calculation methods) are stored using the frame-type database structure used in knowledge engineering. Therefore, the validity of the interrelationship between a plurality of observation results is verified by comparing them with predictions based on the design database, and the failure location is estimated from the contradictory relationship between the observations and predictions. Therefore, creating a database of design information and how to use it will eliminate design drawings without specialized knowledge, and will be useful in investigating the cause of abnormality diagnosis and deciding on countermeasures, and searching by computer will eliminate mistakes. Extremely efficient.

First, creating a database of design information according to the present invention will be explained. In the present invention, when creating a database of design information, (1) As a basic database used for diagnosis, one that corresponds one-to-one with the design data of the plant is used.

(2) Procedural knowledge regarding how to use design data is also included as data and can be used for diagnosis.

This method is used.

For example, the design of the control system and safety system of a power plant is typically based on the following design information.

(a) Expanded connection diagram showing the tag names of each electronic instrument used in the system and their mutual connection relationships.

(b) Connection diagram Name and connection destination of each cable in the system (terminal number of meter)
The one that shows.

(C) Instrument specification clothing Shows the type of each instrument and setting parameter values.

(d) Instrument instruction manual This explains the purpose of the specifications and how to determine input/output relationships for each type of instrument.

In order to materialize such design information as a database, the present invention utilizes knowledge frame representation used in knowledge engineering. As shown in FIG. 2, the frame representation stores data in the form of frame names, attribute names, and attribute values.

Relational databases, the most commonly used database storage method, store data in pairs of attribute names and attribute values, and frame representation has a structure very similar to this.

On the other hand, the advantages of frame representation for relational databases are the concept of ``inheritance of properties'' and the concept of ``demon procedure.''

``Inheritance of properties'' means that a frame above a certain frame is defined, and lower frames can inherit the data of the upper frame.With such a hierarchical structure, the necessary Data can be stored efficiently.

Another advantage is the "demon procedure." This is a procedure that is automatically activated when an attribute value of a certain frame is required, and is used to obtain the necessary data. As shown in Figure 2, when a normal value is requested (if-need
d) when the value is added to the frame (if-adde
d), when the value is removed from the frame (if −r
There are procedures that are invoked in each of the following steps. The ability to store such procedures within the database means that the usage of data can be stored, making it a further extension of conventional relational databases.

Hereinafter, a method for expressing a frame of design information according to the present invention will be explained with reference to a specific example.

As already shown, the design information for the control system and safety system includes ■ Instrument tag names and parameter settings ■ Connection information for each instrument ■ Purpose of instrument use, input/output related functions, input/output port types (handling information) (equivalent to a manual). These are not independent pieces of information, but rather interrelated pieces of information. For example, even though each instrument has a different name, the same type of instrument is often used, and its general characteristics are described in ① above.

This is an example of hierarchical data representation. In addition, regarding the input/output relationship calculation function of the instrument mentioned in (2) above, it is difficult to express it as data, and it is necessary to express it as a procedure (algorithm). Therefore, creating a database of such information requires frame representation.

FIG. 3 shows a concrete example of frame representation.

First, an instrument frame is provided for each instrument in a control system or safety system. Here 1nheritance
(Inheritance of properties of upper frame), lineage name.

The common name of the instrument is required. Setting parameter values are required for each individual instrument. Regarding how to use this parameter value, it is usually necessary to read the instruction manual for each type of instrument, but in order to express this as data, an instruction manual frame is provided. This instruction frame stores general information about the instrument. For example, as an input/output related calculation procedure, a procedure for calculating an output from setting parameter values of individual instruments and input values is written as a demonstration procedure. Also, since the type of input/output port does not depend on the individual instrument, it can be written within this frame. On the other hand, a line frame is provided for the cable connection of each instrument. This line frame indicates which boat and which post a certain cable should be attached to.

By storing data in the form described above, necessary information can be retrieved in various forms. In particular, since there is a one-to-one correspondence with design drawings, information that matches the actual hardware can be retrieved, making detailed use possible.

Next, the principle of the diagnostic method of the present invention using the above-mentioned design database will be explained.

FIG. 4 is a diagram for explaining the basic diagnostic method of the present invention.

Among the instruments in the system that are subject to diagnosis, the main ones are equipped with indicators and recorders. In addition, there are a plurality of observable points such as monitoring points using a Kl calculator and locations where voltage can be measured using a destar. In the diagnostic method according to the present invention, an abnormality is detected and diagnosed from the mutually contradictory relationship of these plurality of observed values (voltage values). That is, as shown in FIG. 4, the observed value S1
When S3 is predicted using the input/output relationship calculation procedure of the design database mentioned above, and it is different from the actual measurement of 83, the device (Tag-1) between the two is predicted.

Tau-2, rag-:3) can be presumed to have an abnormality (determination 1).

82 is similarly predicted from 81, and if the predicted value and the actual value for S2 match, the equipment in between (Tag-
1, Tag-2) can be considered normal (Judgment 2
).

By proceeding with such consistency determination using arbitrary combinations of a plurality of observed signals, it is possible to distinguish between normal equipment and equipment that is likely to be abnormal (comprehensive determination).

The diagnostic method described above is a method of narrowing down devices that may be abnormal using a fault location estimation method using the elimination method. In order to automatically perform this kind of filtering,
In the present invention, the design database is searched in the following manner.

■ Decide on the signal (eg 33) you want to predict.

■ Find out what needs to be observed to predict S3. In the above example, the device whose output is S3 (TA
G-3> input (identified from the input/output port name on the instruction manual frame), and then connect the output 32 (line frame) of the other device (Tag-2) connected by cable from that input. By repeating this step,
You can reach observable locations.

■ Once the observation points necessary for prediction are obtained through the above steps,
The target signal is predicted by finding values in the input->output sequence according to the input-output relationship calculation procedure in the instruction manual frame.

To determine whether or not it is observable using the procedure (2) above, refer to the instrument frame and first search for easily observable points such as indicators and recorders. Also, if this fails, a more time-consuming process such as finding an observation point that can be measured with a tester will be required. Such level division can reduce unnecessary measurements.

On the other hand, regarding the narrowing down procedure described above, in principle it is sufficient to examine all possible combinations between multiple observed values, but in order to increase efficiency, we take the following steps. . In other words, if a discrepancy occurs in a certain combination, search for a spot that can be observed with the equipment between the clear observation points. Furthermore, if they match, the observed value on the input side used for prediction is set as the target, and the observed value for predicting this is searched upstream. Such a procedure makes it possible to narrow down devices that may have abnormalities based on observation data.

In the above implementation procedure, consider the voltage value as observation data,
Since the input/output calculation procedure uses voltage as input/output calculation, this diagnostic method is mainly suitable for abnormal events such as drift events.

On the other hand, if an event such as an alarm 0N10FF or an interlock 0N10FF is input as an observed value and a logical operation is prepared as an input/output measurement procedure, it becomes possible to diagnose malfunctions in the interlock system. Also, as an observed value,
If you use qualitative states (increase, decrease, upscale, downscale) as input/output measurement procedures, and include causal relationships between these qualitative states (for example, if the input increases, the output will increase), then It is also possible to diagnose transient events where voltage values cannot be read.

(Example) The present invention will be explained with reference to the drawings.

FIG. 1 is a flowchart showing one embodiment of the present invention. In this flowchart, numerals 1 to 3 are frame-type databases that store design data. That is, 1
is an instrument frame database, 2 is a line frame database, and 3 is an instruction manual frame database.

Furthermore, a frame search device 4 is provided for searching these frame type databases. Here, the frame search device 4 searches for the value of the attribute name of the requested frame. If there is no corresponding attribute name, go to search the upper frame and search for its value. Further, when a demon procedure is issued, the demon procedure execution device 5 is used to execute the procedure.

On the other hand, the flow of diagnosis is shown in 6 to 11. First, in the first step 6, the observation Y-ta, which is currently 1q at 14 points, is input. In the next second step 7, a signal to be verified for the presence or absence of a contradiction is set from among the signals. Furthermore, in the third step 8, inputs necessary for verification are estimated using frame data. If the required input is not observed, return to the first step 6 and request observation. Next, in the fourth step 9, the observed data in the third step 8 is used while referring to the predicted frame data of the signal set in the second step 7, and in the fifth step 10, the normality is compared with the actual measured value. or abnormality. This procedure is repeated at the first step 6 to the fifth step 10 until narrowing down is possible, and the results are displayed at the sixth step 11. In addition to the above diagnostic procedure, it is also possible to search for frame data using the search request 122 display 13.

As described above, according to this embodiment, the design information of electronic equipment is stored as a frame-type database in which multiple pieces of instrument frame data, line frame data, and instruction manual frame data are hierarchically related to each other.
Since this is used for diagnosis, it has the following excellent effects.

(1) Various events can be handled by preparing input calculation procedures according to the form of observation data.

(2) Even if a set of observation data in a fixed form is not given, it is possible to narrow down the results to the extent possible from only the obtained data. Furthermore, it is possible to handle cases where observed values are added later.

(3) Since it is based on data that directly corresponds to design information, it is possible to diagnose in accordance with the actual hardware. Also,
It can be used not only for diagnosis but also as a design information search system.

[Effects of the Invention] As explained above, according to the present invention, design information of electronic devices is stored as a frame-type database configured to be hierarchically related to each other, and this is used for diagnosis. It has the excellent effect of being able to carry out plant liability diagnosis in a flexible manner that can respond to various plant conditions, and in a detailed manner.

[Brief explanation of drawings]

Fig. 1 shows a 70-diameter diagram according to an embodiment of the present invention, Fig. 2 shows a frame representation according to the invention, Fig. 3 shows design information in a frame representation, and Fig. 4 shows a diagram according to the invention It is a diagram that explains the principle of related diagnosis. 1...Separate device frame database 2...Line frame database 3...Instruction manual frame database 4...Frame search device 5...Demon procedure execution device (8733) Agent Patent attorney Yoshiaki Inomata ( fool
1 person)

Claims (2)

[Claims] (1) Handling the design information of the control system and safety electronic equipment of the power plant, including instrument frame data representing information such as instrument names and setting parameter values, line frame data expressing the connection relationship of instruments, and instrument functions. Explanation frame data is stored as a frame-type database configured to be hierarchically related to each other, and is used to store arbitrary observed values among multiple observed values obtained from the control system and safety system of the power plant. The observed value of the input required for prediction is searched from the frame-type database, and the output of the device is predicted from the observed value of the input by input/output calculation processing of the frame-type database, and this predicted value and the observed value are A plant diagnosis method characterized in that the plant diagnosis method is configured to compare values to determine whether or not there is an abnormality in the equipment. (2) The plant diagnosis method according to claim 1, characterized in that abnormal equipment is narrowed down by repeatedly performing equipment abnormality processing.