JP2783631B2 - Plant abnormality diagnosis device - Google Patents

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial application field) The present invention relates to a plant abnormality diagnosis device having an improved man-machine interface when performing a plant abnormality diagnosis.

(Prior Art) When a plant becomes abnormal, a plant abnormality diagnosis apparatus for diagnosing the cause of the abnormality by using the plant as an object to be diagnosed and deriving a plan for solving the problem is shown in FIG. This will be described with reference to the conventional example.

As shown in FIG. 9, the conventional plant abnormality diagnosis apparatus always uses plant data 2 which is a state value of a plant at a predetermined cycle from a plant 1 to be diagnosed.
A data collection function 3 for inputting the data and a check on the input plant data 2 with a predetermined limit value, and when it is determined that the data is abnormal,
The plant data discriminating function 5 for notifying the plant state 4 of the plant that is abnormal and the main operating state of the plant 1 at the time of being judged as abnormal, for example, the plant state 4 indicating the number of operations, output, etc. An inference engine 7 which refers to the notified plant state 4 and a knowledge database 6 which is a set of knowledge for solving the problem, and derives a plan for solving the problem by taking in the insufficient information in solving the problem; A knowledge database 6 which is a set of knowledge accumulated when the inference engine 7 derives a plan for solving a problem;
A knowledge database input unit 9 for assisting an expert 8 with specialized knowledge in inputting into the knowledge database 6; and an inference engine when the inference engine 7 takes in insufficient information for guiding a solution to a problem. According to the question type 10 that is the keyword of the missing information notified from 7,
A question display function 14 for selecting a question database 12 which is a set of messages for asking the operator 11 performing a diagnosis and displaying the corresponding messages on the input / output device 13, and a question for the operator 11 corresponding to the room type 10 The question database 12 which is a set of messages presented as a question type, and the question type which is a keyword of the insufficient information notified from the inference engine 7 when the inference engine 7 takes in the insufficient information to guide the solution to the problem. Graphic format data which is a set of graphic data of a fixed part of the display screen of the plant data 2 corresponding to 10
15 and plant data 2 corresponding to the question type 10 is received from the data collection function 3, and the plant data display function 16 is displayed on the input / output device 13 so as to be superimposed on the graphic format data 15. A fixed part of the display screen of the corresponding plant data, for example, graphic format data 15 which is a set of graphic data such as scales, units, etc .; a man-machine interface unit 17 comprising these blocks 12, 14, 15, 16; An input / output device 13 for receiving the output from the unit 17 and the main force from the expert 8 of the operator 11
Consists of

The plant abnormality diagnosis device 18 configured as described above is a system for elucidating the cause of the abnormality of the plant 1 and deriving a plan for solving the problem when the plant 1 becomes abnormal. An expert 8 having specialized knowledge or a problem-solving ability similar thereto.

However, the plant abnormality diagnosis device 18 does not perform all the processing in the system to derive a solution to the problem. The inference engine 7 refers to the plant state 4 and the knowledge database 6 which are referred to in the process of inference to derive the solution to the problem. Information that is lacking in the plant,
Therefore, it is difficult to make a determination, and for the information to be determined by the operator 11 performing the diagnosis, a question and the plant data 2 related to the question are presented through the man-machine interface unit 17, and the operator 11 converts the plant data 2 into the question. The inference is advanced by taking in the insufficient information by judging while referring to and inputting the answer. When the inference engine 7 takes in the insufficient information, the question type 10 is notified to the question display function 14 and the plant data display function 16. Question display function
14 searches the question database 12 for a question message corresponding to the notified question type 10 and displays it on the input / output device 13. Further, the plant data display function 16 searches for graphic format data corresponding to the question type 10 and receives the plant data 2 corresponding to the question type 10 from the data collection function 3 so as to overlap the graphic format data 15 with the input / output device. Display at 13. FIG. 10 shows a specific display example.

In the example of FIG. 10, the plant data screen 19 displaying the plant data 2 when the plant becomes abnormal is compared with the plant data 2 during the normal operation of the plant 1 and the result is shown. This is an example in which a question is displayed by displaying a message on the question screen 20 as to whether or not the question is asked. When the operator answers such a question, the operator must first investigate the normal plant data 2 and compare it with the value. At this time, when the normal plant data 2 to be compared is not more constant than the value of the other plant data 2, when the comparison is not constant due to the value of the other plant data 2 at the time of abnormality, the comparison is performed. At this time, the other plant data 2 at the time of the abnormality is also investigated, and the comparison cannot be performed unless the conditions of the other plant data 2 at the time of the abnormality and the normal time are matched. In the example of FIG. 10, since the vibration amplitude value at the time of startup is not a constant value depending on the rotation speed, it is necessary to investigate the vibration amplitude value at each rotation speed at normal time and the rotation speed at the time of abnormality. . As described above, when a question is asked, the operator 11 investigates various data, examines it, and inputs an answer. Also, by repeating the presentation of such a question and the input of the answer, the plant 1
It is possible to clarify the cause of the abnormality when it becomes abnormal and derive a plan to solve the problem.

(Problem to be Solved by the Invention) In the above-mentioned conventional plant abnormality diagnosis apparatus, information that is insufficient for inference, that is, plant data or the like which is determined by an operator because it is difficult to make a determination in the computer is taken into the computer. Therefore, an inquiry is made as a question, but generally, there is no operator who can judge the content of the question and immediately input it to the computer as an answer, and at the time the question is issued, the operator makes a number of current state values in the plant, or a past state value. After examining the state value, etc., it is determined, and the question is answered. When an abnormality occurs in the plant, the cause of the abnormality is determined by inference, and the operation operation instruction to the plant is presented by the man-machine interface based on the cause of the abnormality. In some cases, it may be difficult, and there is a problem that, when the result is hurried, an incorrect answer is input in some cases and an inference result different from the actual result is derived.

Therefore, in the present invention, in the inference process of the plant abnormality diagnosis, when capturing the missing information, instead of presenting only the question and the plant data corresponding to the question, a graphic serving as a determination criterion when determining the content of the question is provided. The data is displayed in accordance with the plant operation status and the contents of the question at the time of occurrence of the abnormality, and the graphic of the plant data to be judged is easily displayed by displaying the data corresponding to the content of the question and the plant operation status. It is an object of the present invention to provide a plant abnormality diagnosis device capable of answering the above.

[Configuration of the Invention] (Means for Solving the Problems) A plant data discriminating function for checking plant data input based on a predetermined limit value and discriminating an abnormality, and a plant abnormality inference engine is provided with a problem of a plant abnormality. A question display function that displays questions on the display screen of the input / output device to capture information that is missing from the plant status and knowledge database when deriving a solution, and a graphic display of plant data that is input A plant abnormality diagnosis device having a plant data display function and an input / output device that indicates the main operating state of the plant at the time when the plant data notified from the plant data determination function corresponding to the type of the question is determined to be abnormal. Select appropriate plant data and graphic format data to be displayed on Display data discriminating means for notifying the plant data display function, and when selecting the plant data and graphic format data to be displayed on the input / output device, the display data discriminating means determines the type of the question, the state of the plant and the input / output device. The plant data to be displayed in the query plant data correspondence database, which is a set of data associated with graphic format data, and the question display function displays the question data on the input / output device, and at the same time, notifies the display data discriminating means. Based on the plant data and the graphic format data, reference data determining means for creating reference data for determining the content of the question and displaying the data on the input / output device, and the reference data determining means determine the content of the question When creating data that serves as the basis for A reference value database which is a set of necessary reference value data is added.

(Operation) In the above configuration, the display data discriminating means receives the type of the question from the inference engine and the state of the plant determined by the plant data discriminating function, and compares the two types of information with the plant data to be displayed. Refers to the question plant data correspondence database in which the correspondence table is defined, finds the plant data to be used for determining the question by pattern matching, and identifies the identification codes of the plant data and graphic format data that are correlated with the plant data, and displays the plant data. Pass to function and reference data discriminating means. The plant data display function superimposes the corresponding graphic format data and the plant data used for determining the question on the input / output device. The reference data determining means receives the plant data used for determining the question, the plant data correlated with the plant data and the identification code of the graphic format data, and receives the correlated plant data from the data collection function. Next, based on the data, the plant data having a correlation in the reference database and the reference data used for determining the question are interpolated using the reference data used to determine the question, and the data serving as the determination reference corresponding to the state value of the plant is created. Display on the I / O device.

(Embodiment) FIG. 1 is a block diagram showing a plant abnormality diagnosis apparatus according to an embodiment of the present invention. In the figure, the same reference numerals as those in FIG. 9 indicate the same or corresponding parts, and the point different from the configuration in FIG. 9 is that when the inference engine 7 derives a solution to the problem, the state 4 of the plant and the knowledge database 6 are used. In the process of performing inference while referring to the information of the plant state 4 and the information lacking in the knowledge database 6, that is, the plant abnormality diagnosis device
Since it is difficult to judge by only 18, the question display function 14 displays the data of the question database 12 on the input / output device 13 at the same time as incorporating the information to be judged by the operator 11 who makes the diagnosis, The appropriate plant data 2 is displayed on the input / output device 13 based on the plant state 4 representing the main operating state of the plant 1 at the time when the plant data 2 notified from 10 and the plant data discriminating function 5 is determined to be abnormal. Therefore, the plant data 2 to be displayed and the graphic format database
Display data discriminating means 20 for selecting 15 data and notifying the data to the reference data discriminating means 19 and the plant data display function 16
When the display data discriminating means 20 selects the plant data 2 and the data of the graphic format database 15 to be displayed on the input / output device 13, the question type 10,
The question plant data correspondence database 21 which is a set of data in which the plant state 4 and the plant data 2 to be displayed on the input / output device 13 and the graphic format data are associated with each other. At the same time as displaying on the input / output device 13, the reference data database 22, which is a set of data serving as a reference for determining the content of the question, based on the plant data 2 and the graphic format data notified by the display data determination means 20, Reference data discriminating means 19 for selecting data in the graphic format database 15 and creating reference data for judging the content of the question and displaying the data on the input / output device 13; Standards required when creating data to be used as criteria for judging the contents of The point is that a data display system 23 including a reference value database 22 which is a set of value data is added.

With the above configuration, the plant data 2 which is the state value of the plant 1 is input from the plant 1 to be diagnosed by the data collection function 3 at a predetermined cycle at all times. Further, a check is performed by the plant data discriminating function 5 based on a predetermined limit value. If the data is determined to be abnormal, the inference engine 7
Then, the state 4 of the plant at that time and the state 4 of the plant determined to be abnormal are notified. The inference engine 7 executes inference with reference to the notified plant state 4 and the data of the knowledge database 6. At this time, if the information to be determined is insufficient, the inference engine 7 temporarily suspends the inference here, and the question display function 14 and the display data discriminating means 20 determine the type 10 of the question as a keyword of the information lacking. Notify. The question display function 14 extracts a question message corresponding to the question type 10 notified from the inference engine 7 from the question database 12 and displays the question message on the input / output device 13. on the other hand,
The display data determining means also determines the plant data 2 and the graphic format data to be displayed based on the question type 10 notified from the inference engine 7 and the plant state 4 notified from the plant data determining function 5.

Hereinafter, a specific example of the format of the query plant data correspondence database 21 used to determine the display data of FIG. 2, the query plant data correspondence data of FIG. 3, and the processing flow of the display data determination means 20 of FIG. It is explained along.

First, the database corresponding to the question plant data in Fig. 2
21 formats will be described. The data is one set in one line, and the question type 10 notified from the inference engine 7 and the plant state 4 notified from the plant data determination function 5 and the question type 10 used for pattern matching,
The plant status 4 and the data ID 124, which is the identification code assigned to the plant data 2 used for determining the question, the data ID 225, which is the identification code assigned to the plant data correlated with the data ID 124, and the display Screen and reference part (scale, unit, etc.) of plant data to be changed
And graphic format data which is an identification code of

Next, each item of the question plant data correspondence database 21 will be described according to a specific example of the question plant data correspondence database 21 shown in FIG.

The question type 10 is a keyword that simply represents the content of the data displayed on the input / output device 13. In the first line of the specific example, an SV diagram showing the correlation between the amplitude value of the first bearing and the rotation speed is shown. It means that it is data. The state 4 of the plant is a predetermined symbol such as "1" if the rotation speed is increasing, and "-1" if the displayed plant data graphic is data that does not depend on the state 4 of the plant. data
The ID 124 corresponds to data to be output to the input / output device 13 and is a predetermined symbol. Data ID2 25 is Data ID1
The same as 24, but the same data is used to display a graphic indicating the correlation with other plant data 2 depending on the plant state 4. For example, even when judging the change state of the amplitude value, if the rotation speed is increasing, this is used to display an SV diagram showing the correlation between the rotation speed and the amplitude value. Graphic format data 15 is data ID 1 2
4. The graphic (scale, scale) of the reference part when displaying the plant data 2 corresponding to the data ID 225 on the input / output device 13
In actual display, the plant data 2 is displayed so as to overlap the graphic format data 15 corresponding to the graphic format ID 26.

Next, the processing flow of the display data discriminating means 20 in FIG. 4 will be described.

The display data discriminating means 20 reads one line of the question type 10, the plant state 4, and the question plant data correspondence data (401, 402). The question type 10 is compared with the question type 10 in the question plant data correspondence database 21 to see if it is the same (403). If not, the pointer of the question plant data correspondence database 21 is advanced by one line (404), and the question is asked again. One line is read from the plant data correspondence database 21 (40
2) Repeat the same process until the pattern matches.
If the pattern matches (Yes in 403), the plant state 4 and the plant state 4 of the query plant data corresponding data
If it is not the same, the pointer of the question plant data correspondence database 21 is advanced by one line (404), the question plant data correspondence data is read again by one line (402), and the question type 10 Is performed. This process is continued until both the question type 10 and the plant state 4 match. When both patterns match, the data ID 124, the data ID 225, and the graphic format ID 26 are read (406) and notified to the plant data display function 16 and the reference data discriminating means 19 (407).

The plant data display function 16 includes a display data determination unit 20.
Receiving the plant data 2 corresponding to the data ID 1 24 and the data ID 2 25 notified from the data collection function 3,
Further, a data file in which the graphic format data corresponding to the graphic format ID 26 is stored is searched from the storage device, the graphic format data is displayed, and the plant data 2 is displayed thereon.

The reference data discriminating means 19 receives the data ID 1 24, data ID 2 25, graphic format ID from the display data discriminating means 20.
26, and the data serving as the criterion are displayed on the input / output device 13 by these. Hereinafter, a reference value database used to create the data serving as the determination reference in FIG.
The description will be made in accordance with the format of No. 22 and the processing flow of the reference data discriminating means in FIG.

First, the format of the reference value database 22 shown in FIG. 5 will be described. The data is one set in one row, and is used for determining the question corresponding to the reference value correlation data 27 and the reference value correlation data 27 which is data correlated with the plant data 2 used for determining the question. The reference value data 28 is used as a reference for determining the plant data 2. These data are used as data for obtaining the reference value data 28 corresponding to the data correlated with the plant data 2 used for determining the question received from the data collection function 3 by interpolation. The reference value correlation data 27 is stored in the reference value database 22 in ascending order of value.

Next, an explanation will be given according to the processing flow of the reference data discriminating means 19 in FIG.

The reference data discriminating means 19 receives the data ID 1 24, data ID 2 25, graphic format ID from the display data discriminating means 20.
26 received (601), graphic format ID 2
The graphic format data 26 corresponding to 6 is displayed on the input / output device 13 (602), and then the plant data 2 corresponding to the data ID 225 is received and read into the buffer (60).
3). This data is hereinafter referred to as plant data 2m. m
Indicates a variable indicating the point stored in the buffer.

Next, one line of data is read from the reference value database 22 (604). The read data is the reference value correlation data 27m,
This is referred to as reference value data 28m. n indicates a pointer in the reference value database 22.

Then, the plant data 2m read as described above is compared with the reference value correlation data 27n (605).
(606). If larger, reference value database 2
The pointer of 2 is advanced by one line (607), and then the data pointed to by the pointer is read (608). Next, the read reference value correlation data 27n is compared with the plant data 2m (609). As a result, if the plant data 2m is large, the process is repeated again from the process of advancing the reference value and the pointer of the database 22 (607). When the plant data 2m is small, the reference value data 28m corresponding to the plant data 2m is interpolated by the reference value correlation data 27n and the reference value data 28n and the data stored in the pointer immediately before the data. Approximately obtained (610). Next, the obtained reference value data 28m is displayed on the input / output device 13 (60
6). The above processing is performed for all the plant data 2m read into the buffer, and it is determined whether or not the processing is completed (61).
1) If not completed, the pointer of the plant data 2m read into the buffer is advanced by one (612), and the process 604
Repeat again from In this manner, data serving as a criterion for the plant data 2 to be determined is created and displayed on the input / output device 13.

FIG. 7 shows a question screen 29 displaying an inquiry database 12 on the input / output device 13 when taking in the information that the inference engine 7 lacks, and a plant data 2 when the plant 1 becomes abnormal. 13 shows a display example of a plant data screen 30 displayed on 13 and a reference screen 31 in which data serving as a criterion for determining plant data 2 when the plant 1 becomes abnormal are displayed on the input / output device 13. In this display example, the vibration amplitude value at the time of abnormality detection at the time of startup is compared with the value of the vibration amplitude value at the normal time, and the result is determined. When the operator 11 answers the question displayed on the question screen 29, he understands the contents of the question screen 29 and compares the plant data 2 displayed on the plant data screen 30 with the reference value data 28 displayed on the reference screen 31. By comparing the contents, it is possible to judge on the spot and input an answer. In this way, in the inference process, when the operator 11 is inquired of the missing information, the plant data 2 adapted to the state at the time of occurrence of the abnormality of the plant 1 corresponding to the question is used as the question and the judgment standard of the question. By displaying the plant data on the input / output device 13, the operator 11 can easily input an answer without investigating a large number of plant data 2 in the plant 1. In addition, the input of the answer to the question displays data serving as a criterion, so that there is no variation in the value to be answered depending on the level of the operator 11, the degree of experience, and the like, and the diagnosis result does not differ depending on the operator. .

The range of the value of the reference value correlation data 27 in the reference value database 22 in FIG. 5 is assumed to be all integers, and is correlated with the plant data 2 used for determining the question passed from the data collection function 3 to the reference data determination means 20. By converting the related plant data 2 into integers, the reference data discriminating means 20 can approximately create data serving as a judgment reference.

FIG. 8 shows a processing flow of the reference data discriminating means 19 according to another embodiment of the present invention.

The reference data discriminating means 19 is provided by the display data discriminating means 20,
Data ID 1 24, Data ID 2 25, Graphic format
Receives ID 26 (801), graphic format ID
The graphic format data corresponding to 26 is displayed on the input / output device 13 (802), and then the plant data 2 corresponding to the data ID 225 is transmitted and read into the buffer (80).
3). Next, change this data to integer type data (80
Four). Hereinafter, this data is referred to as plant data 2i. i
Indicates a variable indicating the pointer stored in the buffer.

Next, one line is read from the reference value database 22 (80
Five). The read data is referred to as reference value correlation data 27n and reference value data 28n. n indicates a pointer in the reference value database 22.

It is compared whether or not the plant data 2i converted into the integer and the read reference value correlation data 22n are the same (806). If they are not the same, the pointer of the reference value database 22 is advanced by one line (807), and the reference value database 22 is read (805).
Start with. If they are the same, the reference value data 28n is input / output device 1
Displayed on 3 (808). Next, it is determined whether the data of the plant data 2i is the last row (809). If not, the pointer of the plant data 2i is advanced by one (81).
0), the process starts with the process of reading the reference value database 22 (805). By repeating this process, data serving as a criterion in the plant data 2 to be determined can be approximately created and displayed on the input / output device 13.

[Effects of the Invention] As described above, according to the present invention, in the inference process of the expert system, when an operator asks the operator for information that is lacking in the knowledge base and obtains an answer, instead of presenting only the question, it is not necessary to present only the question. By presenting the plant data at the time of the abnormality and the plant data as the criterion at the same time to assist the question determination adapted to the content and the state at the time of the plant abnormality, the operator can input the answer to the question at present or in the past. Answers can be input easily without investigating a large number of state values of the plant. Also, since the input data of the answer to the question is determined based on the data serving as a criterion, there is variation due to the level of the operator, the degree of experience, and the like. Without this, the same diagnostic result can be obtained.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an expert system showing one embodiment of the present invention, FIG. 2 is a format diagram of a query plant data correspondence database of FIG. 1, and FIG. 3 is a specific example of a query plant data correspondence database of FIG. FIG. 4 is a processing flowchart of the display data discriminating means of FIG. 1, FIG. 5 is a format diagram of a reference value database of FIG. 1, and FIG. 6 is a reference data of FIG. FIG. 7 is an explanatory diagram showing a display example of a question screen, a plant screen, and a reference screen displayed on the display screen of the input / output device of FIG. 1, and FIG. 8 is another drawing of the present invention. FIG. 9 is a processing flow diagram of the reference data discriminating means in the embodiment, FIG. 9 is a configuration diagram of a conventional device, and FIG. 10 is an explanatory diagram showing a display example of a conventional question screen and a reference screen. 19: Reference data discriminating means, 20: Display data discriminating means, 21: Questionable plant data correspondence database, 22 ...
… Reference value database, 23 …… Data display system.

──────────────────────────────────────────────────続 き Continued on the front page (58) Fields surveyed (Int. Cl. ⁶ , DB name) G01D 21/00 G06F 9/44 G05B 13/02

Claims (1)

(57) [Claims] 1. A data collection function for constantly inputting plant data at a predetermined cycle from a plant to be diagnosed, and the input plant data is checked by a predetermined limit value and determined to be abnormal. A plant data discriminating function for notifying a plant state of an abnormal condition, a knowledge database which is a set of knowledge data for solving a problem and a plant state value notified from the plant data discriminating function, and this knowledge database. An inference engine that refers to and derives a plan for solving a problem by incorporating information that is insufficient for solving the problem, an input / output device having a screen display and a keyboard for inputting and outputting necessary data, and the knowledge database Data to input expert knowledge of the plant to be diagnosed from outside Source input unit, when the inference engine that performs the dialogue with the operator takes in the information that is insufficient for solving the problem, the information that is insufficient is determined by the type of the question that is the keyword of the information that is informed by the inference engine. A question database which is a set of messages for asking an operator performing diagnosis, a question display function for referring to the question database and displaying a corresponding message on the input / output device, and the inference engine is insufficient for problem solving. At the time of incorporating the information, the plant data input through the data collection function for determining the content of the message displayed on the input / output device by the question display function, a set of graphic data in a fixed portion of a data display screen Plan to be displayed on the input / output device over the status display format data Data display function, a graphic format database that is a set of graphic data of a fixed part of the plant data display screen corresponding to the type of the question, when the inference engine derives a solution to the problem, the plant state, the knowledge base The main data of the plant at the time when the plant data notified from the plant data discriminating function corresponding to the type of the question for taking in the missing information in the knowledge base in the process of performing the inference while referring to the plant is determined to be abnormal. Display data discriminating means for selecting appropriate plant data and the graphic format data for displaying the operating state on the input / output device, and notifying the plant data display function; and the display data discriminating means displays on the input / output device. Plant data and graphic format data When selecting, the type of the question, plant data to be displayed on the input / output device and the state of the plant, a question plant data correspondence database that is a set of data associated with graphic format data, and the question display function At the same time as displaying the question data in the input and output, by the plant data and the graphic format data notified by the display data determination means,
Reference data determining means for creating reference data for determining the content of the question and displaying the data on an input / output device; and when the reference data determining means creates data serving as a reference for determining the content of the question. And a reference value database which is a set of reference value data required for the plant.