JPH06124209A - Knowledge base construction supporting device - Google Patents

Abstract

PURPOSE:To integrate the correct knowledge base of a sufficient amount, to develop an operation supporting device with high reliability and to improve the development efficiency of the operation supporting device. CONSTITUTION:A term dictionary editing module 1 controls the editing of a term dictionary for supporting knowledge base development and a term registration module 2 registers new terms inputted from an input/output device 5 to a term dictionary to be tentatively stored in a memory 3 corresponding to an instruction from the module 1 or term classification in a prescribed order. A term processing module 4 performs the retrieval, comparison and reference recording, etc., of the inputted terms corresponding to the instruction from the module 1 by referring to the term dictionary inside the memory 3. A term dictionary management module 6 is provided with a function for reading the term dictionary preserved in a preserving device 7 to be tentatively stored in the memory 3, the function for displaying tentatively stored contents from the memory 3 and the function for arranging the term dictionary tentatively stored in the memory 3 to be written to the preserving device 7.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a knowledge base construction support device applied to a knowledge base construction for an operation support device for a thermal power plant or the like.

[0002]

2. Description of the Related Art As a method for supporting the operation of plants and machinery, there is a method of making an expert system by using a knowledge engineering method. The operation support expert system is a system that incorporates knowledge and know-how possessed by experts regarding the operation of plants and machinery as a knowledge base, and allows a computer to make judgments similar to those of a human expert.

An example of such a driving support expert system is shown in FIG. In FIG. 8, the part surrounded by a thick broken line is the driving support device 200, and the part surrounded by a double line is a part of the driving support expert system 300.

(I) The data collecting device 210 receives pressure / pressure from a plant or machinery 100 that is the target of operation support.
Process quantities such as temperature, control signals such as alarm signals, trip signals, valve opening command signals, and equipment state quantities such as valve opening and pump start / stop status (these are collectively referred to as "plant parameters" below) To collect. (ii) The driving support expert system 300 includes a state grasping module 310, a cause identifying module 320, and a countermeasure determining module 330.

(Iii) The state grasping module 310 uses the state grasping knowledge base 311 to detect an abnormality sign occurring in the plant, provides the plant state information 312 to the operator, and at the same time extracts a candidate of an abnormality cause. To do.

(Iv) The cause diagnosis module 320 uses the cause diagnosis knowledge base 321 to diagnose the cause of the detected abnormal sign, provides the cause diagnosis information 322 to the operator, and at the same time proposes a countermeasure. Extract.

(V) The countermeasure determination module 330 uses the countermeasure determination knowledge base 331 to determine a countermeasure in accordance with the detected abnormality sign or the diagnosed abnormality cause, and provides the operator with the operation guide 332.

(Vi) As shown in the flow of processing in FIG. 8, the data collecting device 210 collects "plant parameters", the state grasping module 310 grasps the plant state, the cause diagnosing module 320 diagnoses the cause of abnormality,
And the measure determination by the measure determination module 330 is
It is repeatedly performed at regular intervals.

(Vii) The state grasp knowledge base 311 is used for knowledge of abnormal signs which requires monitoring and response of operators,
And a knowledge base that stores knowledge for describing meta-knowledge for extracting candidate abnormal causes to be investigated when this abnormal sign is detected. FIG. 9 shows an example of the description format of the state grasp knowledge, and FIG. 10 shows an example of the knowledge described according to this description format.

In the following, following the general terminology of knowledge engineering,
The entire description of FIG. 9 is called a frame, particularly an abnormal frame in this case, (trigger condition ...) Inside one stage is called a slot, and (trigger condition name) + inside another stage is called a slot value. Also, in the figure, you can describe any number of one or more with + after it, indicating that there is an or relationship to each other, and those with # in front should be described only when necessary Show. The lower broken line indicates that the term is fixed for each driving support device. The contents described in FIG. 9 are as follows. (1) Abnormal frame name Phenomenon name showing the abnormal sign of the plant whose status is to be grasped. (2) Trigger condition slot

The condition (trigger condition) for detecting this abnormal sign will be described. Any number can be specified and they are considered to be mutually or conditions. That is, when even one is established, the cause frame given to the next cause frame slot is extracted as a cause candidate. The trigger condition must be a binarized variable. For example, it is an alarm signal installed in the plant. (3) Cause frame slot When this abnormal sign is detected, the name of a cause frame described below that can be a cause candidate is given. The example in FIG.
(1) An abnormal symptom that occurs in the plant called "reheat steam temperature low abnormality" (2) Pre-alarm "reheat steam temperature control deviation-
It is the knowledge to detect by the occurrence of "Large" or the transmission of the plant alarm "Reheat steam temperature caution", and (3) The cause frame "RH heat transfer function abnormality" is extracted as a cause candidate.

In order to input the knowledge example shown in FIG. 10 into a computer, a panel-type frame editor may be used. An example of the input panel screen of the abnormal frame is shown in FIG.
Shown in.

(Viii) The cause diagnosis knowledge base 321 is a knowledge base having a hierarchical structure corresponding to the functional configuration of the diagnosis target, and can describe the knowledge element for diagnosing the abnormality of each function. FIG. 12 shows an example of a hierarchical structure of a cause diagnosis knowledge base taking a thermal power plant as an example.

In FIG. 12, the output operation function abnormality, which is the function abnormality of the entire thermal power plant to be diagnosed, is set as the highest level, and the primary factors that cause this function abnormality are the furnace output function abnormality, the in-core combustion function abnormality, Six of the heavy water supply function abnormalities, heat transfer function abnormalities, ventilation system functional abnormalities, and electric energy conversion abnormalities indicated by the bold lines in the figure are the one lower-order functional abnormalities, and sub-functional abnormalities that cause each of these functional abnormalities (for example, RH heat transfer function abnormality, SH heat transfer function abnormality, etc.) as a sub-function of heat transfer function abnormality)
Finally, it has a hierarchical structure in which a functional abnormality developed to a desired fineness, such as a functional abnormality of an individual device (eg, GR control system abnormality, GR fan abnormality, GR tamper abnormality) is the lowest.

A graphic input type network editor may be used to input the hierarchical structure shown in FIG. 12 to a computer. An example of the input screen of the network editor is shown in FIG. In the figure, a rectangular frame for defining a cause frame described later that the network editor has as an input environment component, an arrow indicating the upper / lower relationship, or a Japanese input sub-key for inputting the cause frame name in the rectangular frame. It is an example of a screen during network construction using a window. A description format example of the cause diagnosis knowledge is shown in FIG.
An example of knowledge described according to this description format is shown in FIG.
Hereinafter, the entire description of FIG. 14 will be referred to as a cause frame.
The contents described in FIG. 14 are as follows. (1) Cause frame name The name of the functional abnormality to be diagnosed. (2) Upper slot Describe the upper function. For the slot value, the upper cause frame name, the judgment procedure, and the local variable name are (name (fun arg) v
ar) format.

The name is the name of the higher cause frame (observation event name in (3)), and the proof certainty factor of the higher cause frame (in (3), occurrence rate of observation event) is entered as a value in the variable of the same name. ing. (fun arg) permits the following in the procedure for determining the satisfaction of conditions. (G n1): value of name is greater than n1 (L n2): value of name is less than n1 (B n1 n2): value of name is between n1 and n2 (T): value of name is true ( NIL): The value of name is false var is the name of the local variable that is valid only within this cause frame, and the value is the true / false value of the determination result of (fun arg). (3) Proof condition slot Describe each condition necessary for proving the cause frame. For the slot value, any combination of observation event name, decision procedure and local variable name is described in the form of (name (fun arg) var). name, (fun arg) and var are the same as above (2)
It is as described in the section of the upper slot of. (4) Confidence factor calculation slot The description format of this slot is (confidence factor calculation (condition confidence factor local variable name) +)

The conditions for determining the certainty factor for establishing the cause candidate, the certainty factor, and the local variable name are given. condition is,
It is described in and / or logic expression using the local variable defined in (2) or (3) or the plant parameter having a true / false binary value. The certainty factor is "0.0 ≤ certainty factor ≤ 1.0, in accordance with criteria determined by the application system developer such as physical certainty as a cause, empirical / subjective certainty, ease of countermeasures / maintenance, etc. Given in the range of. The local variable name is the name of a local variable that is valid only within this cause frame, and the value is true if the condition is satisfied and false if the condition is not satisfied or has not been evaluated. Any number of (condition certainty factor local variable name) pairs can be given, but shortcut processing is performed. The shortcut processing is to judge the conditions (assuming whether the condition is true or false) for the slot value written in the form of (condition xxx) +, and if any one condition is true, This is a process of processing the part of xxx and ignoring all the (condition xxx) below it. Therefore, when describing conditions, it is necessary to start with strict conditions.
When all the conditions are not satisfied, the certainty factor of the cause frame is set to "0.0". (5) Countermeasure candidate slot

A countermeasure frame name, which will be described later, describing countermeasures for this cause candidate is given. However, the countermeasure frame given to this slot is extracted as a countermeasure candidate only when the certainty factor of the cause frame becomes positive. (6) Lower slot

When this cause candidate is proved, the cause candidate to be proved is described as a more detailed cause with a condition. Any combination of conditions and any number of sub-cause frames can be given, but shortcut processing is performed. Figure 15
For example, (1) "RH heat transfer function abnormality" is an abnormal cause candidate, and (2) if the credibility of the proof of the upper frame "heat transfer function abnormality" exceeds "0.8". The local variable U1 that is valid only within the frame is set to T. (3) As proof conditions, the following six plant parameters are examined. (1) If TRHO control automatic is true, set the local variable P1 to T
To (2) If the reheat steam flow abnormality is false, set the local variable P2 to T
To (3) If TRHO low is true, set local variable P3 to T
To (4) If the load fluctuation is false, set the local variable P4 to T
To (5) If the TFSO height is true, set the local variable C1 to T
To (6) If TRHI low is false, set the local variable C2 to T
To (4) The method of determining the certainty factor is (a) (and U1 P1 P2 P3 P4 (or C1 C
If 2)) is true, the certainty factor is set to "0.8" and the local variable F1 is set to true. (b) (and U1 P1 P2 P3 (or C1 C2))
Is true, the confidence factor is set to "0.5" and the local variable F2 is set to true. (c) If (and U1 P1 P2 P3) is true, the certainty factor is set to "0.2" and the local variable F3 is set to true. (5) If the proof certainty factor of this cause frame has a positive value, the countermeasure frame “RH heat transfer function abnormality countermeasure” is extracted as a countermeasure candidate. (6) After the verification of the cause frame, if (a) (and F1 C2) is true, the lower cause frame “IR control abnormality” is set. (B) If only F1 is true, the lower cause frame “GR control abnormality” is set. Is extracted as the next cause candidate. Is an example.

In order to input the knowledge example shown in FIG. 15 into a computer, a panel-type frame editor may be used. Input panel screen example of the cause frame is shown in FIG.
Shown in.

(Ix) The countermeasure decision knowledge base 331 is a knowledge base that stores knowledge for the operator to decide the countermeasure to be taken according to the detected abnormality candidate and the diagnosed abnormality cause. FIG. 17 shows an example of the description format of the countermeasure decision knowledge, and FIG. 18 shows an example of the knowledge described according to this description format. Hereinafter, the entire description of FIG. 17 will be referred to as a countermeasure frame. The contents described in FIG. 17 are as follows. (1) Countermeasure frame name The name of a candidate for countermeasures such as operation. (2) Countermeasure candidate slot

An arbitrary set of a plurality of operation guides is described as a condition for selecting operation candidates. The condition is based on the processing result of the status grasp module or the cause diagnosis module.
Describe with and / or / not logic expression. What is an operation guide?
It is a guidance message issued from the output device. The process of combining the condition and the operation guide is a shortcut process. In the example of FIG. 18, (1) the countermeasure candidate “RH heat transfer function abnormality countermeasure” is (2) as the countermeasure (a) if the proof certainty of the RH heat transfer function abnormality is larger than “0.7”, "Reduce the load to the in-house load", "disconnect", and make the "warm state" state or later.

(B) If the proof certainty factor of the RH heat transfer function abnormality is larger than "0.4" and the proof certainty factor of the SH heat transfer function abnormality is larger than "0.4", "load up to internal load""Descent" and then "enhance monitoring" to identify the abnormal progress. (C) If the proof certainty of the RH heat transfer function abnormality is greater than “0.2”, “supervised monitoring” is performed to determine the subsequent abnormal development situation. It should be noted that a panel-type frame editor may be used to input data to the computer shown in FIG. FIG. 19 shows an example of the input panel screen of the countermeasure frame.

The performance of the driving support expert system exemplified above is determined by the goodness of the knowledge base. Conventionally, knowledge base development requires knowledge and knowledge possessed by experts regarding plant and machinery operation. A method has been adopted in which a knowledge engineer extracts and organizes know-how through interviews and the like to build a knowledge base. However, even in the frame representation illustrated here,
Or in other knowledge representation formats such as rule representation,
In general, an inference program essentially executes inference processing only by matching character patterns, so if there is any difference in the observed event name, stored variable name, frame name, etc., it is treated as another. Knowledge engineers were responsible for preventing such errors and for finding and correcting errors.

[0025]

[Problems to be Solved by the Invention]

(1) Since the knowledge engineer was completely left to unify the terms for the knowledge representation, there was a possibility of making an error of using multiple representations for one event or frame and including the error in the constructed knowledge base. there were.

(2) In order to unify the terms, there is a problem that a great deal of fatigue is required such as creating a term dictionary on paper in advance or searching and referring to the term dictionary and previously created knowledge data one by one when inputting the knowledge base. .

Due to these factors, the development efficiency of the driving support device is remarkably reduced, and a sufficient amount of correct knowledge base cannot be incorporated, or the driving support device provides erroneous diagnosis results and operation guides, and the plant trip There is a risk of causing a loss such as leading to.

The present invention has been made in consideration of the above situation, and it is possible to incorporate a sufficient amount of a correct knowledge base, which makes it possible to develop a highly reliable driving support device and improve its development efficiency. It is an object of the present invention to provide a knowledge base construction support device that can contribute to the improvement of plant availability and safety.

[0029]

DISCLOSURE OF THE INVENTION The present invention is a knowledge base for a driving support device mainly composed of a driving support expert system for grasping the state of a plant or machinery which is the target of driving support, identifying the cause and determining a countermeasure. In a device that supports construction work, a term dictionary editing means for controlling the term dictionary for the target plant or machinery and the entered terms are registered in the term dictionary in a predetermined order according to the term classification. The term registration means for storing the term dictionary, the storage means for temporarily storing the term dictionary in accordance with the term classification, and the term dictionary in which the entered term is registered are referred to for processing such as search, comparison and reference recording. Term processing means for performing, input / output means for inputting data or request, outputting question or term dictionary, and term dictionary management for performing term dictionary management Characterized by comprising the stage, and a storage means for storing the contents of dictionaries.

[0030]

The term dictionary editing means controls the term dictionary for supporting the knowledge base development. The term registration means temporarily stores in the storage means in a predetermined order according to a command from the term dictionary editing means or according to the term classification of new terms input via the input / output means. Register in the existing term dictionary. The term processing means refers to the term dictionary temporarily stored in the storage means for the input term in accordance with a command from the term dictionary editing means, and performs processing such as search, comparison, reference recording and the like.

The term dictionary management means reads the term dictionary saved in the saving means and temporarily stores it in the storage means, displays the contents temporarily stored in the storage means, and temporarily stores in the storage means. It has three term dictionary management functions of organizing the term dictionaries stored in the memory and writing them to the storage means.

With the above configuration, it becomes possible to edit the term dictionary on the computer while constructing the knowledge base. By automatically referring to this term dictionary, 1
It is possible to reduce or prevent errors such as using one event name or multiple expressions, and to unify the terms, create a term dictionary on paper in advance or use the term dictionary and previously created knowledge data one by one when entering the knowledge base. The effort of referencing can be reduced, and the efficiency of knowledge base construction work can be significantly improved.

[0033]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a knowledge base construction support device according to an embodiment of the present invention. In the figure, a computer 10 is hardware for mounting a term dictionary editing module 1, a term registration module 2, a memory 3, a term processing module 4 and a term dictionary management module 6 realized as software, and a data bus 8 The input / output device 5, the storage device 7, and the knowledge base editor 20 are connected via the. The knowledge base editor 20 is implemented on the hardware different from the computer 10 here, but may be installed on the computer 10. The term dictionary editing module 1 is responsible for editing a term dictionary for supporting knowledge base development.

The term registration module 2 responds to a command from the term dictionary editing module 1 or the input / output device 5
The new terms input via the are registered in the term dictionary temporarily stored in the memory 3 in a predetermined order according to the term classification.

The term processing module 4 refers to the term dictionary temporarily stored in the memory 3 in response to a command from the term dictionary editing module 1 to search, compare, record reference, etc. Is processed. I / O device 5
Is a standard CRT display device equipped with a keyboard capable of multi-window display. A knowledge base developer inputs various requests such as data and display requests for building a knowledge base, and a term dictionary editing module. According to the command from 1, the question and the term dictionary are output to the knowledge base developer.

The term dictionary management module 6 includes a storage device 7
The operation of reading out the term dictionary stored in the memory 3 temporarily, storing the term dictionary in the memory 3, displaying the content temporarily stored in the memory 3, and organizing the term dictionary temporarily stored in the memory 3 , Of the action of writing to the storage device 7
It has one term dictionary management function. The storage device 7 is a standard data storage device such as a fixed disk device or a floppy disk device. Hereinafter, the functions and actions of the above-mentioned parts will be described in more detail.

The term dictionary editing module 1 registers the observed event name, the storage variable name, the abnormal frame name, the cause frame name or the countermeasure frame name for creating the abnormal frame, the cause frame or the countermeasure frame described in the conventional technique. -A module corresponding to the main program that controls the flow of various processes such as search and display, and has four arguments.

In the following, the observation event name, the storage variable name, the abnormal frame name, the cause frame name and the countermeasure frame name are collectively referred to as “term”, and which is which is referred to as “term classification”. Also, when it is necessary to distinguish the individual terms and term classifications that have been input as processing targets from the terms and term classifications that have already been registered in the term dictionary, the terms "input term" and "input term classification" As described above, the two characters "input" are added and called. The first argument is an argument that specifies the usage of this device, and is represented by the following symbols for convenience of description. A: Term dictionary editing mode: This device is activated independently and only term dictionary editing is performed. B: Knowledge base construction mode: Activated from the knowledge base editor 20 to edit the term dictionary while constructing the knowledge base. The second argument is an argument indicating a command of processing to the knowledge base construction support device, and here, for convenience of description, it is represented by the following numerical value. 1: Initial setting of term dictionary 2: Display of term dictionary 3: Search of term 4: Registration of term 8: Saving term dictionary 9: Termination Third argument and fourth argument are different for each process, and will be described individually. . The processing flow of the term dictionary editing module is shown in FIG. (1) Initial setting of term dictionary

The contents of the command are checked by steps A1 to A6, and when "command = 1", the third argument is the name given to the term dictionary stored in the storage device 7, for example, the file name (in the following, a simple Therefore, it is called a “term dictionary file”) and the fourth argument is not used. The process calls the term dictionary management module 6 described later to initialize the memory 3 (step A7). (2) Display of term dictionary

When "command = 2", the third argument is the name of the term classification to be displayed, and the fourth argument is not used. In the process, the term dictionary management module 6 described later is called to display the relevant area of the term dictionary stored in the memory 3 (step A8). (3) Search for terms

When "command = 3", the third argument is an input term to be processed. The fourth argument is an input term classification, and a plurality of parameters can be specified using array type variables.
In the processing, the term processing module 4 described later is called to search for the input term and record the reference (step A9).
). (4) Term registration

When "command = 4", the third argument is an input term to be processed. The fourth argument is an input term classification, and only one can be specified. The process calls the term registration module 2 described later to register the input term (step A10). (5) Save term dictionary

When "command = 8", the second argument is the term dictionary file name stored in the storage device 7, and the fourth argument is not used. The process calls the term dictionary management module 6 to be described later to cause the term dictionary temporarily stored in the memory 3 to be saved (step A11). (6) Command input

After performing the above processing, only when it is judged in step A12 that "usage = A", that is, the term dictionary editing mode, the command for processing to the knowledge base construction supporting device is input again. Perform (step A13).

Note that the usage is specified by the knowledge base developer or the knowledge base editor 20 when the apparatus is started, and the command and the term classification are, when "usage = A", the knowledge base via the input / output device 5. Specified by the developer,
In the case of "usage = B", the knowledge base editor 20 recognizes and specifies the knowledge base construction status, the type of frame, and the term classification permitted as the slot value.

The term registration module 2 is called from the term dictionary editing module 1 with the usage, command, term classification, and input term as input arguments only in the usage = A, that is, the term dictionary editing mode. 1) Check already registered, (2) Register new term, (3) Read input term,
Is processed. (1) Check already registered

It is checked whether or not the input term is already registered in the term dictionary temporarily stored in the memory 3, and if it is already registered, the input term and the registered term classification are attached and a caution message is input. Output via the output device 5, ask if you want to cancel the registration or re-enter the input term. (A) If you cancel, end the processing of the input term, (b) If you re-enter, (3) ) Jump to the process of reading the input term. (2) Registration of new terms

If the input term is not registered in the term dictionary, the input term is registered in the memory 3 in the order of the Japanese syllabary in a predetermined order according to the term classification. It should be noted that “to register” means to add an input term to the memory 3 with the reference count “0” in a data structure shown below. (3) Reading input terms Input terms are read in the following two cases. In case of re-input of 1 ° (1) (b) 2 ° In case of “usage = A”, when registering a term of the same terminology continuously. Prepare the special terms that mean. A flowchart of the above processing is shown in FIG. First, it is checked whether it is already registered (step B1), and if it is not already registered, the term is registered (step B2).

In the case of already registered, judgment processing of "cancel?", "Re-input?" And "end?" Is performed (step B3.
B5), and if none of them, it is judged whether or not "usage = A" (step B6). Also, step B2
If the term is registered in step B3, or if it is determined in step B3 that the term is canceled, step B6 is performed. Then, in this step B6, "how to use =
If it is judged to be "A", or if it is judged to be re-input at step B4, the term is inputted and then the process returns to step B1.

The memory 3 temporarily stores the term dictionary being edited. As shown in FIG. 3, one term dictionary is divided into five areas of an observation event name area, a storage variable name area, an abnormal frame name area, a cause frame name area, and a countermeasure frame name area. The data structure in Figure 4
As shown in, the data is a set of term and reference count data. The reference count is a numerical value that is set to "0" when the term is registered in the term dictionary and incremented by 1 each time it appears in the knowledge base. The term processing module 4 is called from the term dictionary editing module 1 with the usage, term classification and input term as input arguments, and performs the following processing. (1) Reference record Check whether the input term is already registered in the term dictionary temporarily stored in the memory 3. (a) If it is already registered and the term classification matches,

(A) When "Usage = A", the input term is added to the term classification and the message indicating that the term is already registered is output via the message input / output device 5, and the processing relating to the input term is finished. (B) When "Usage = B", the number of times the term is referred to in the memory 3 is increased by 1, and the process relating to the input term is finished.

(B) If the term classifications do not match in the existing registration, the input term and its term classification are output together with the caution message via the input / output device 5, an overlap error is set, and the next similar term search is performed. move on. (2) Similar term search (c) Collect all similar terms of the input term from the term dictionary.
The term “similar term” is determined in advance by defining a similar term rule such as “only one character is different”. (d) If there is a similar term, all the similar terms and their term classifications are output via the input / output device 5, and the process proceeds to the next new term registration. (3) New term registration (e) If the above (1) (b), that is, duplicate error decoding is set, the processing for the input term is finished. (f) In other cases, ask if you want to register new terms,

(B) In the case of registration, the input terms are registered in the memory 3 in a predetermined order such as the Japanese syllabary order and the reference frequency is "1" in accordance with the term classification. The reference count is set to "1" not only to indicate that the term has been registered, but also that the term has been referenced once. Or, if you can't narrow down to one term classification, ask the knowledge base developer. (B) When not registering, the processing regarding the input term ends. A flowchart of the above processing is shown in FIG.

The term dictionary management module 6 is called from the term dictionary editing module 1 with the command and the term dictionary file name (when the command is 1 or 8) / term classification (when the command is 2) as an input argument. Is processed. (1) Initial setting of term dictionary When the command is 1, (a) When the term dictionary file name is given, (a) The term dictionary is read from the term dictionary file saved in the storage device 7 and stored in the memory. Set to 3,

(B) Ask the knowledge base developer whether to reset the reference record, and only when resetting, reset the reference count of all the terms in the memory 3 to "0", and finish the term dictionary management process. (b) When the term dictionary file name is not given, the contents of the memory 3 are emptied. (2) Display of term dictionary When the command is 2, the term classification applicable part of the term dictionary stored in the memory 3 is displayed via the input / output device 5. (3) Saving term dictionary When the command is 8,

(A) Ask the knowledge base developer whether or not to organize the term dictionary in the memory 3. Only when organizing the term dictionary, all the terms in the memory 3 are referred to as "0". Delete the reference count set data. (b) The term dictionary temporarily stored in the memory 3 is written in the term dictionary file on the storage device 7. FIG. 7 shows a flowchart of the above processing.

[0058]

As described in detail above, according to the present invention, it is possible to edit a term dictionary directly on a computer or to edit a term dictionary on a computer while building a knowledge base. 1 by automatically referring to the dictionary
Errors such as using one event name or multiple expressions can be reduced or prevented. Also, for unification of terms,
A great deal of labor such as creating a term dictionary on paper in advance or referring to the term dictionary or previously created knowledge data one by one when inputting the knowledge base is greatly reduced, and the efficiency of the knowledge base construction work can be significantly improved. Furthermore, a term dictionary for device development can be used, and the effect of improving the reliability of the knowledge base by unifying the terms between the operation support device knowledge bases for the same type plant is achieved.

As described above, a sufficient amount of a correct knowledge base can be incorporated, a highly reliable driving support device can be developed, and the development efficiency of the driving support device can be remarkably improved. It also has an effect of contributing to the improvement of safety.

[Brief description of drawings]

FIG. 1 is a block diagram showing a knowledge base construction support device as an embodiment of the present invention.

FIG. 2 is a flowchart showing the processing of a term dictionary editing module 1.

FIG. 3 is an explanatory interactive diagram of area division of a term dictionary in the memory 3.

FIG. 4 is a diagram showing a data structure in each area of a term dictionary in the memory 3.

FIG. 5 is a flowchart of processing of a term registration module 2.

FIG. 6 is a flowchart of processing of a term processing module 4.

FIG. 7 is a flowchart of processing of a term dictionary management module 6.

FIG. 8 is an explanatory diagram of a processing module configuration and a processing flow of the driving support device.

FIG. 9 is a diagram showing a description format example of state grasp knowledge.

FIG. 10 is a diagram showing an example of state grasp knowledge described according to the description format of FIG. 9.

FIG. 11 is a diagram showing an example of an abnormal frame input panel screen.

FIG. 12 is a diagram showing an example of a hierarchical structure of a cause diagnosis knowledge base taking a thermal power plant as an example.

FIG. 13 is a diagram showing a screen example during construction of a cause frame network diagram by a network editor.

FIG. 14 is a diagram showing a description format example of cause diagnosis knowledge.

15 is a diagram showing an example of cause diagnosis knowledge described according to the description format of FIG.

FIG. 16 is a diagram showing an example of a cause frame input panel screen.

FIG. 17 is a diagram showing a description format example of countermeasure decision knowledge.

FIG. 18 is a diagram showing an example of countermeasure decision knowledge described according to the description format of FIG. 17;

FIG. 19 is a diagram showing an example of an input panel screen of a countermeasure frame.

[Explanation of symbols]

1 ... Term dictionary editing module, 2 ... Term registration module, 3 ... Memory, 4 ... Term processing module,
5 ... I / O device, 6 ... Term dictionary management module, 7 ... Storage device, 8 ... Data bus, 10
…calculator.

Claims (6)

[Claims] 1. A device for supporting a knowledge base construction work for a driving support device, which is mainly composed of a driving support expert system for grasping the state of a plant or machinery which is the target of driving support, identifying the cause, and determining a countermeasure, A term dictionary editing means for controlling the editing of the term dictionary relating to the target plant or machinery, and a term registering means for registering the input term in the term dictionary in a predetermined order according to the term classification, Storage means for temporarily storing the term dictionary according to the term classification; term processing means for performing processing such as search, comparison, reference recording, etc. by referring to the term dictionary in which the entered term is registered, Input / output means for inputting data or request and outputting a question or term dictionary; term dictionary managing means for managing term dictionary;
A knowledge base construction support device comprising a storage unit for storing the contents of a term dictionary. 2. In the term dictionary editing means, a term dictionary editing mode in which the present apparatus is independently activated to perform only term dictionary editing, or knowledge which is activated from a knowledge base editor or the like to edit a term dictionary while building a knowledge base. When one of the usage modes of base construction support mode is specified, control the flow of each process of initializing the term dictionary, displaying the term dictionary, searching for terms, registering terms, saving and ending term dictionary. The knowledge base construction support apparatus according to claim 1, wherein 3. The term registration means, when used in the term dictionary editing mode, checks whether or not the input term is already registered in the term dictionary temporarily stored in the storage means, and checks whether the term is already registered. Then, a caution message is output, and if not registered, input terms are registered in a predetermined order in a term dictionary temporarily stored in the storage means in accordance with the term classification, according to claim 1. Knowledge base construction support device. 4. A term dictionary in which the storage means is divided into areas corresponding to a predetermined term classification, and the data structure in each area is a set of term and reference frequency group data. 2. The knowledge base construction support device according to claim 1, wherein the knowledge base construction support device is temporarily stored. 5. The term processing means, when the term dictionary editing means calls the usage, term classification and input terms as input arguments, the terms entered are temporarily stored in the storage means. Check if it is already registered in the dictionary, and if it is already registered in the term dictionary edit mode and the term classification matches, a means for ending the process related to the input term, and already registered and term classification in the knowledge base construction support mode. If the two match, the means for increasing the reference count of the term in the term dictionary temporarily stored in the storage means by “1” and ending the process relating to the input term; If does not match,
A means to output a caution message and set a duplicate error code, and regardless of the usage mode, if the input term is not registered, apply a predetermined similar term rule and display all similar terms, The knowledge base construction support apparatus according to claim 1, further comprising means for asking a user whether to register as a new term and performing registration or termination processing. 6. The knowledge base construction support according to claim 1, wherein the term dictionary management means initializes the term dictionary, displays the term dictionary, or saves the term dictionary based on a command from a user. apparatus.