JPH02293632A - Plant diagnostic device - Google Patents

Abstract

PURPOSE:To decrease an operator's burden by providing a means for deciding the change tendency of plant data and executing the diagnosis processing of an expert system by using the result of the decision. CONSTITUTION:A discontinuous point detecting means 26 starts a disturbing point detecting means 27 when there is the discontinuous point in the plant data which detects abnormality. The means 27 announces the generation of the disturbing point to the expert system 9 when the means discriminates the generation of this point. The means start a sudden change detecting means 28 when the disturbing point cannot be detected. The means 28 announces the sudden change to the system 9 upon detection of the generation of the sudden change. On the other hand, the means 26 starts a gentle change detecting means 29 when the discontinuous point is not detected. The means 29 announces the detection of the gentle change to the system 9 upon detection of the generation of such change. The means 29 starts a rapid change detecting means 30 when the means does not detect the gentle change. The means 30 investigates the change rate of the plant data and announces the rapid change in the change tendency to the system 9 upon decision of such change.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention is a plan 1 for diagnosing the cause of an abnormality occurring in a plant using a rotating body abnormality diagnosis expert system.
・Regarding diagnostic equipment. (Prior Art) For example, a thermal power generation unit uses a large number of so-called rotating bodies, such as a main turbine, turbines such as a boiler feed pump turbine, and pumps such as a boiler feed pump. The stability of plant operation can be improved by early detection of abnormalities in these rotating bodies, investigation of the cause, and countermeasures.

In particular, in recent years, thermal power plants have been using DDS (Daily Start) to adjust the power load.
p and Shutdown) or W S S (W
Eekly Startup and Shutdown
) Operation has become widespread, and as a result, starting/stopping operations are also performed frequently, resulting in frequent temperature changes in the rotating parts of the rotating body, making it more likely that abnormalities will occur in the rotating body. Therefore, in order to prevent serious accidents at thermal power plants, when an abnormality occurs in a rotating body. Investigate the cause of the abnormality;
A plant diagnosis device using a rotating body abnormality diagnosis expert system is in practical use as a device for diagnosing and outputting countermeasures, and a conventional example thereof is shown in FIG. The plant data is collected at regular intervals, and the collected plant data is stored in the plant data storage means 3 in chronological order. As shown in FIG. 12, the plant data storage means 3 stores in chronological order no plant data necessary for data collection time and abnormality cause estimation for each collection operation of the plant data collection means 1. At the beginning of the storage area , the latest data collection time representing the time of the latest data collection operation performed by the plant data collection means 1 is stored. The data table 4 for abnormality detection determination is the first
As shown in FIG. 3, for each plant data, a limit value for determining the value of each plant data as abnormal is stored.

As shown in FIG. 14, the abnormality detection data name table 5 stores, for each plant data, an abnormality detection data name that represents the abnormality content in a kanji code string when each plant data is determined to be abnormal. It is something.

As shown in FIG. 15, the data table 6 for collecting plant data around the abnormality diagnosis time includes data before abnormality detection that represents the number of samplings before the time of abnormality detection, out of the number of samplings required for determining the change trend of plant data. It stores the number of samplings to be collected, m, and the number of samplings to be collected after anomaly detection, m, which represents the number of samples to be collected after detecting an anomaly. The abnormality detection determination means 6 reads the latest data collection time in the plant data storage means 3, finds a data collection time that matches the latest data collection time, and then
The plant data is compared with the limit values stored in the data table 4 for abnormality detection and judgment to determine which plant data is in an abnormal state, and when any abnormality is detected, the abnormality is The detected plant data is notified to the abnormality detection time peripheral plant data collection means 8, and the system waits for a time until the collection operation of the abnormality detection time peripheral plant data collection means 8 is completed, and when the waiting period is completed, the expert system 9 The inference engine 91 is activated, and at the same time, the anomaly detection data name corresponding to the plant data in which an anomaly is detected at that time is extracted from the anomaly detection data name table 5 and output to the external procedure interface 92 of the expert system 9.

When the abnormality detection time surrounding plant data collecting means 8 is notified of the abnormality detected plant data by the abnormality detection determining means 7, the abnormality detection time surrounding plant data collection means 8 stores the data in the abnormality diagnosis time surrounding plant data collection data table 6 with that point in time as a reference. The sample values of the notified plant data are inputted from the plant data storage means 3 as many as the pre-abnormality detection data collection sampling number II11, and each sample value and its collection time are combined into a set for diagnosis. It is stored in the plant data storage means 10, and from that point on, the notified plant data sample values are inputted from the plant data storage means 3 as many times as the number m2 of data collection samplings after abnormality detection, and each sample is stored. The value and its collection time are stored as a set in the diagnostic plant data storage means 1o.

As a result, as shown in FIG. 16, the diagnostic plan 1/data storage means 10 stores positive sample values before the detection time of the plant data detected as an abnormality by the abnormality detection/determination means 7, and II12 sample values after the time are stored in pairs with their respective collection times.
Note that in this case, the detection time is time Q. Further, time T represents the sampling period of the plant data collection means 1.

When the diagnostic plant data output means l1 is notified of an output command from the external procedure interface 92 of the expert system 9, the diagnostic plant data output means 11 displays time on the horizontal axis and time on the vertical axis based on the stored contents of the diagnostic plant data storage means 10 at that time. The printer 12 forms print information of a line graph of each plant data, and outputs the print information to the printer l2 to print out the graph.

The plant data change trend survey content display means 13 forms display information for displaying the inquiry contents when inquiry information about plant data change trends is inputted from the external procedure interface 92, and the display information is displayed on the display operation device. 14, and the display operation device l4
The output is displayed on a display device.

The plant data change trend investigation result input means 15 notifies the external procedure interface 92 of the result information when the inquiry result of the plant data change trend is inputted from the display/operation device 14 .

The inference result display means 16 is an external procedure interface 92
When the expert system 9 is notified of the inference result performed by the expert system 9, display information of the inference result is formed, and the display information is output to the display operation device 14, whereby the inference result is displayed on the display operation device 14. is displayed.

In the expert system 9, the knowledge Heath 93 is
The fact data table 94 stores the production rules processed by the inference engine 91, and the fact data table 94 stores the fact data to which the inference engine 9l applies the production rules. Further, the external procedure interface 92 is for the expert system 9 to exchange necessary information with the outside.

With the above configuration, for example, when the abnormality detection and determination means 7 detects that the first bearing vibration amplitude value, which is the plant data with the order number 2 at time RO, has become abnormal, the abnormality detection and determination means 7 The sequence number 2 of the first bearing vibration amplitude value is notified to the plant data collection means 8 around the abnormality detection time as identification information of the plant data in which the abnormality has occurred, and the system waits until time m and T have elapsed from that point.

Then, at the time (Q+a+,T) at which the anomaly detection determination means 7 finishes its standby, it starts up the inference engine 91, and at the same time starts the "first A Kanji code string called "Bearing Vibration Book Value" is output to the external procedure interface 92.

On the other hand, the plant data collection means 8 around the abnormality detection time,
When the abnormality detection and determination means 7 notifies the sequence number of the abnormality occurrence plant data, the collection times from time (Q-miT) to time (1)+m2T) are sequentially found from the plant data storage means 3, and the collection times are The plant data (namely, the first bearing vibration amplitude value) stored in the second storage area is sequentially extracted and sequentially stored in the diagnostic plant data storage means 10. However, the collection times from time (11+T) to time (Q+n+,T) are sequentially processed when those times are reached.

When the external procedure interface 92 inputs the abnormality detection data name "first bearing vibration amplitude value" from the abnormality detection determination means 7, it registers the abnormality detection data name in the fact data table 94 in the following format. (Abnormality detection data ゜Data name 1st bearing vibration amplitude value)...(A) The inference engine 91 is proceeding with inference by pattern matching of the production rules stored in the fact data table 94 and the knowledge base 93. .

Here, the production rule consists of a condition part and an execution part each having one or more conditional statements and executable statements, and has the following format.

(Condition part) → (Execution part) Note that the details of the production rule are well known in the art, so a description thereof will be omitted.

In this case, in the knowledge base 91, (Abnormality detection data 'Data name 1st bearing vibration amplitude value)
. . (A) → The conditional statement (A) of the production rule and the data (A) of the fact data table 94 match.

Therefore, in this case, the execution statement (c) is first executed, and the external procedure interface 92 instructs the diagnostic plant data output means 11 to display plant time series data. 11 is the time (Q+n+,T) stored in the diagnostic plant data storage means 10 from the time (Q−+niT).
The "first bearing vibration amplitude values" up to this point are taken out, a line graph is formed with the horizontal axis set to time and the vertical axis set to the vibration amplitude value, and the graph is recorded and output from the printer 12.

Furthermore, when the executable statement (engineering) is executed, the external procedure interface 92 causes the plant data change trend investigation content display means 13 to display a kanji code that reads "What is the change trend of the first bearing vibration amplitude value?" as display information. The message display is commanded with the column specified, so that
The plant data change trend investigation content display means 13 displays on the display operation device 14 an inquiry message such as "What is the change trend of the first bearing vibration amplitude value?"

In this way, when the inquiry message is displayed on the display/operation device 4, the operator can look at the line graph showing the change in the first bearing vibration amplitude value recorded and output from the printer 12, and check the trend of the change. , determine whether it corresponds to a "random point,""suddenchange,""gentlechange," or "sudden change."

For example, when it is determined that a "random score" has occurred, the display/operation device 14 is operated to input a kanji character string "random score".

In this way, when a kanji string is input from the display operation device 14, the plant data change trend survey result input means 15
notifies the external procedure interface 92 of the kanji code string representing the content. External procedure interface 92
In this case, notifies the inference engine 91 of the kanji code string input from the plant data change trend investigation result input means 15 as the return value of the argument <ans) of the executable sentence (engineering).

Since the return value of the argument <ans> is obtained, the inference engine 91 executes the executable statement (e), and as a result, the external procedure interface 92 receives the following information as information regarding the change trend of the first bearing vibration amplitude value. fact data table 9
4. Register the following data.

(1st bearing vibration amplitude value 'Change trend Random points)... (Force) As a result, the conditional statement (K) and conditional statement (Q) in the condition part of the production rule of knowledge base 93, and the fact data table. 94 data (
A) and data (force) are matched, and the conditional part of this production rule is established.

Therefore, in this case, the external procedure interface 92 infers an inference result message consisting of a Kanji code string "The cause is unstable vibration" and a countermeasure output message consisting of a Kanji code string "Please adjust the r bearing oil pressure." The result is output to the result display means 16.

As a result, the inference result display means 16 outputs an inference result message saying "The cause is unstable vibration" and a countermeasure output message saying "Please adjust the r bearing oil pressure" to the display operation device 14 in respective predetermined formats. do. In this way, the contents of the abnormality that has occurred in the plant 2, the cause of the abnormality, and the contents of countermeasures are displayed on the display/operation device 14, so that the operator can perform work while referring to the contents. (Problem to be Solved by the Invention) However, with such conventional devices, the operator is required to detect abnormalities, judge the change trend of plant data, and input the results, which requires the skill of the operator who made the judgment. The results of determining the change tendency of plant data may vary depending on the degree of sensitivity and subjectivity, which may lead to variations in the diagnostic results.

In addition, when an abnormality occurs in a plant during plant operation and it is necessary to investigate the cause on the spot, complex judgments such as determining trends in changes in plant data over time are required in such emergencies. This increases the burden on the operator, and there is a risk that it may interfere with driving operations.

An object of the present invention is to provide a plant monitoring device that can solve the problems of the conventional devices and significantly reduce the burden on the operator.

[Structure of the Invention] (Means for Solving the Problems) The present invention provides a diagnostic plant data storage means for storing diagnostic plant data obtained by collecting plant data in a predetermined range before and after an abnormality is detected; Linear approximation processing means for calculating, by linear approximation, an estimated value for determining a plant data change tendency based on the memory contents of the diagnostic plant data storage means, and a diagnosis for storing the estimated value calculated by this linear approximation processing means. a linear approximation processing parameter storage means for storing the linear approximation processing parameters calculated by the linear approximation processing means; and storage contents of the diagnostic plant data storage means and the diagnostic plant estimation value storage means when an abnormality is detected. a discontinuous point detecting means for detecting the occurrence of a discontinuous point in the plant data based on the discontinuous point, and a diagnostic plant data storage means that is activated when the discontinuous point detecting means detects a discontinuous point. and random point detection means for detecting the occurrence of random points in the plant data based on the memory contents of the diagnostic plant estimated value storage means, and notifying the rotating body abnormality diagnosis expert system when a random point is detected. When this random point detection means does not detect any random points, it is activated and detects that a sudden change has occurred in the plant data based on the stored contents of the diagnostic plant estimated value storage means and the linear approximation processing parameter storage means. and a sudden change detection means that notifies the rotating body abnormality diagnosis expert system when a sudden change is detected, and a linear approximation processing parameter storage means that is activated when the discontinuity point detection means does not detect a discontinuity point. A gentle change detection means that detects that a gentle change is occurring in plant data based on the content and notifies the rotating body abnormality diagnosis expert system when a gentle change is detected, and this gentle change detection. It is activated when the means does not detect a gradual change, and detects that a sudden change has occurred in the plant data based on the stored contents of the linear approximation processing parameter storage means.When a sudden change is detected, the rotating body abnormality diagnosis expert It is equipped with a sudden change detection means that notifies the system.

(Function) Therefore, since it is determined whether the change tendency of the abnormal plant data corresponds to a random point, a sudden change, a gentle change, or a sudden change, there is no need for the operator to perform such determination work. , that burden will be reduced. In addition, since variations in the determination results of the plan 1 data change tendency are significantly suppressed, variations in the diagnosis results of the rotating body abnormality diagnosis expert system are also eliminated.

(Embodiments) Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 shows a plant diagnosis device according to an embodiment of the present invention. In this figure, the same parts and corresponding parts as in FIG. 11 are given the same reference numerals.

The linear approximation processing means 21 removes disturbance elements included in the plant data stored in the diagnostic plant data storage means 10, so that a change tendency determination of the plant data, which will be described later, can be appropriately performed. It is used to form data, and is activated when the plant data collection means 8 around the abnormality detection time completes its plant data collection operation.

The data table 22 for linear approximation processing stores the number n (n<-, n<a,) of sampling data for linear approximation processing used by the linear approximation processing means 21 in its processing.

Now, the linear approximation processing means 21 linearly approximates the plant data stored in the diagnostic plant data storage means 10 in the following manner.

In the following explanation, time (4-m1T) is replaced with time T, time (Q+m,T) is replaced with time (T, +nrT),
Further, it is assumed that an abnormality in the plant data yJ is detected at time (d.÷jT) (0≦j≦nr).

The linear approximation processing means 2l uses the least squares method and calculates the following equation (1
), an estimated value yj of yJ is calculated using n sampling data.

y4 = a40nT+bj (n-1≦j≦nr)
- (1) Here, the parameter aJ+l)j is expressed as follows.

In this way, the estimated value yJ calculated by the linear approximation processing means 21 is added with the corresponding time (TO"jT),
It is stored in the diagnostic plant estimated value storage means 23 in the format shown in FIG. 2, and the parameter aJ is linearly approximated in the format shown in FIG. 3 with the corresponding time (TQ"jT) added. It is stored in the processing parameter storage means 24.

The data table 25 for investigating plant data change trends is as follows:
Various data necessary for the change trend determination process performed by the discontinuous point detection means 26, random point detection means 27, sudden change detection means 28, gentle change detection means 29, and sudden change detection means 30 are stored for each plant data. An example of the stored contents is shown in FIG.

The lower limit value P0 for discontinuous point determination is determined by the discontinuous point detection means 26,
The upper limit for random point determination [pz, the lower limit width P for random point determination, and the number of sampling data for random point determination 07 are the random point detection means 27
The lower limit value P4 of the estimated value change rate for sudden change determination and the upper limit value P4 of the estimated value change rate for sudden change determination are determined by the sudden change detection means 28, and the lower limit value P6 of the parameter maximum value for gradual change determination and the upper limit value P6 of the parameter maximum value for gradual change determination. The maximum value upper limit value P7 of the parameter for use is used by the gradual change detection means 29, and the lower limit value P of the maximum value of the parameter for sudden change determination is used by the sudden change detection means 30, respectively.

FIG. 5 shows discontinuous point detection means 26, random point detection means 27,
The outline of plant data change trend determination processing by the sudden change detection means 28, the gentle change detection means 29, and the sudden change detection means 30 is shown.

When automatic data change trend determination is commanded from the external procedure interface 92 of the Sibert system 9, the discontinuity point detection means 26 detects the abnormality within a certain time range before and after the abnormality detection time.
It is determined whether or not there is a discontinuous point in the plant data in which an abnormality has been detected, and if a discontinuous point is determined, the random point detection means 27 is activated (determination 101).

The random point detection means 27 checks whether a random point has occurred within a certain period of time after the occurrence of a discontinuous point, and when it is determined that a random point has occurred, it externally detects the "random point" as a data change trend. The procedure interface 92 is notified. Furthermore, if the occurrence of random points cannot be detected, the sudden change detection means 28 is activated (judgment 102, process 103).

The sudden change detection means 28 chronologically examines the change trend of the plant data within a certain period of time before and after the occurrence of a discontinuous point, determines whether a sudden change phenomenon is occurring, and determines whether a sudden change phenomenon is occurring. If this is determined, the data change trend will be “sudden”.
is notified to the external procedure interface 92. Furthermore, if the occurrence of a sudden change cannot be detected, the external procedure interface 92 is notified that "analysis is not possible".

On the other hand, if the discontinuous point detection means 26 does not detect a discontinuous point, the gentle change detection means 29 detects the rate of change in the plant data for a certain period of time before and after the abnormality detection time. If the rate of change is within a predetermined range, it is determined that the change trend is a "gentle change." If that determination result is obtained, an external procedure is performed to determine the change trend as a "gentle change." interface 9
Notify 2. Furthermore, if a "gentle change" cannot be determined, the sudden change detection means 30 is activated (judgment 1).
07, processing 108).

The sudden change detection means 30 examines the rate of change in plant data for a certain period of time before and after the abnormality detection time, and when the rate of change is within a predetermined range, determines that the change trend is "sudden change", and the determination result is If it is, the external procedure interface 92 is notified of "sudden change" as the change trend (judgment 1).
09, processing 110). In addition, if it cannot be determined that there is a "sudden change", "Analysis not possible" can be determined using the external procedure interface 9.
Notify 2.

In this way, the discontinuous point detection means 26, the random point detection means 27, the sudden change detection means 28, the gentle change detection means 29
, and the sudden change detection means 30 are activated as appropriate to determine the change tendency of the plant data in which the abnormality has been detected, and the results are input to the expert system 9 via the external procedure interface 92.

The processing contents of the discontinuous point detection means 26 will be explained in detail using FIG. 6. Note that changes in plant data yi are
I? In ILl, a change in the estimated value yi is shown by a broken line L2.

? First, the difference ii (“I
yt-ytl) in the range (n-1)≦j≦n.
Monitoring is performed sequentially starting from i=(n-1). Then, when O<ε■<P1 holds true for all samples within that range, it is determined that there is no discontinuous point, and
The gentle change detection means 29 is activated.

On the other hand, for example, when i=n' and i1>P, it is determined that a discontinuous point exists at that point, and the time (TO+n' T) at which the discontinuous point was determined is determined. In the notified state, the random point detection means 27 is activated.

In this way, the discontinuous point detection means 26 detects the time (T,
+n'T) Estimated value that corrects false positives due to noise expected from the previous n point data and estimates the true value? , a point where the difference from the measured data exceeds a threshold value Pi, and therefore the measured data changes discontinuously beyond the range of false detection due to noise, is determined as a discontinuous point. The processing of the random point detection means 27 will be explained in detail using FIG. In addition, plant data y. The change in is expressed by the broken line L3,
Changes in the estimated value yi are indicated by 1iL4.

The random point detection means 27 detects "(9"1yll'1■) and (yn'+J - yn
'+j-,) have different signs' and 'IyI,'-Yll'-t Dpz≦1yゎ'. J-
Yn'+J-, 1≦1y0'yaj19'yaJ-zl"P
If j of xJ exists with 1≦j≦n7, it is determined that the change tendency of the plant data at this time is a “random point”,
The kanji code string that is a “random point” is sent to the external procedure interface 9.
Output to 2.

Further, the j is 1≦j≦n, . If the odor does not exist, the sudden change detection means 2 is notified of the discontinuity point occurrence time (T, +n'T) notified by the discontinuity detection means 26.
8.

In this way, the random point detection means 27 detects pulse-like discontinuous points, that is, random points, as shown in FIG.

The processing contents of the sudden change detection means 28 will be explained in detail using FIG. 8. In addition, the change in the plant data y is plotted as 5
Then, the estimated value y (n', n''+n) is set to points 1, 6, L
Indicated by 7.

First, from the linear approximation processing parameter table 24, i=
Take out the parameter al when n', n'+n, and use I
Yn'-Yn'++nl>P4, and la++'-a
It is checked whether the relationship n'+ol<Ps is established, and if it is, it is determined that the change trend of the plant data at this time is "sudden", and the kanji code string "sudden" is externally transmitted. Output to the procedure interface 92.

In addition, when the above-mentioned relationship does not hold, it means that the change tendency of the plant data at this time is not a "sudden change", and therefore neither the random point detection means 27 nor the sudden change detection means 28 can determine that " The kanji code string "Unanalyzable" is output to the external procedure interface 92. In this way, the sudden change detection means 28 detects the data change tendency before and after the discontinuous point, as shown in FIG. A similar phenomenon, that is, a sudden change, is detected.

The processing contents of the gentle change detection means 29 will be explained in detail using FIG. 9. Note that the change in the estimated value y is expressed by the broken line L8.
Indicated by

First, from the linear approximation processing parameter storage means 24, (n
-1) Take out the parameter ai in the range of ≦i≦nf,
Find the one with the largest absolute value among them, and set that value as a, .
shall be.

And P,<a. ■＜p. Check whether it holds true. When this relationship holds true, it is determined that the change trend of the plant data at that time is a "gentle change", and in that case, the kanji code string for "gentle change" is sent to the external procedure interface 9.
Output to 2. Further, if the above-mentioned relationship does not hold, the parameter a. ．． 8, the sudden change detection means 30 is activated. The processing contents of the sudden change detection means 30 will be explained in detail using FIG. 10. Note that a change in the estimated value y is shown by a broken line L8.

The sudden change detection means 30 detects the parameter a. ．． ! When received, it is checked whether the relationship amat>p holds. When this relationship is established, the change trend of the plant data at that time is
In that case, the kanji code string "sudden change" is output to the external procedure interface 92.

In addition, when the above-mentioned relationship does not hold, the change trend at this time is not a "sudden change" and, therefore, neither the gentle change detection means 29 nor the sudden change detection means 30 could determine, so it becomes "unanalyzable". The kanji code string is output to the external procedure interface 92. With the above configuration, if the abnormality detection and determination means 7 determines that the first bearing vibration amplitude value is abnormal, the abnormality detection and determination means 7 activates the inference engine 91 of the expert system 9 in the same way as in the conventional device. At the same time as starting up, a Kanji code string "first bearing vibration amplitude value" is output to the external procedure interface means 92 as the abnormality detection data name.

Due to this. The external procedure interface means 92 registers the following data in the fact data table 94. (Abnormality detection data ゜Data name 1st bearing vibration amplitude value)
...(Sa) In this case, in the inference engine 9l, the (abnormality detection data ゜data name 1st bearing vibration amplitude value) of the knowledge base 9l
...(shi) → wag-sand (automatic judgment of data change trends, (a
ns>) +++ (s)make (1st bearing vibration amplitude value 'change trend <ans>) ・
- The conditional statement (shi) of the production rule (ce) and the data (sa) of the fact data table 94 match. Therefore, in this case, the executable statement (S) is first executed, and the external procedure interface 92 instructs the discontinuity point detection means 26 to automatically determine the tendency of data change. As a result, the discontinuous point detecting means 26 is activated, and as described above, the discontinuous point detecting means 26, the random point detecting means 27,
The sudden change detection means 28, the gradual change detection means 29, and the sudden change detection means 30 determine the change tendency of the plant data.

At this time, for example, if a discontinuous point is detected by the discontinuous point detection means 26, the random point detection means 27 is activated, and the change tendency of the plant data is determined by the random point detection means 27 as a "random point", ``Random point'' from the random point detection means 27
A kanji code string is output to the external procedure interface 92.

In this case, the external procedure interface 92 notifies the inference engine 91 of the kanji code string input from the random point detection means 27 as the return value of the argument <ans> of the executable statement (su). Since the return value of the number <an9> is obtained, the executable statement (Se) is executed, and the external procedure interface 92 uses the fact data table 9 as information regarding the change trend of the first bearing vibration amplitude value.
4. Register the following data.

(First bearing vibration amplitude value ゛Change trend Random points) ... (Y) and conditional statement (T) are matched with data (S) and data (T) in the fact data table 94, respectively, and this production rule is The conditional part is satisfied.

Therefore, in this case, the external procedure interface 92 infers an inference result message consisting of a Kanji code string "The cause is unstable vibration" and a countermeasure output message consisting of a Kanji code string "Please adjust the r bearing oil pressure." Output to result display means 16. As a result, the inference result display means 16 outputs an inference result message saying "The cause is unstable vibration" and a countermeasure output message saying "Please adjust the bearing oil pressure" to the display operation device 14 in respective predetermined formats. do. In this way, the cause of the abnormality that has occurred in the plant 2 and the details of the countermeasures are displayed on the display and operation device 14 without any operation by the operator, so the operator can refer to the contents and carry out work. can.

As described above, in this embodiment, when an abnormality occurs in the plant 2, the cause of the abnormality that has occurred and the details of countermeasures are automatically displayed without operator intervention, so the operator is not burdened. is significantly reduced.

In addition, since these displays can be obtained relatively quickly after an abnormality occurs, even when an abnormality that requires immediate action occurs, the operator can take appropriate action with plenty of time. The stability of plant 2 operation is improved.

By the way, in FIG. 1, the parts surrounded by broken lines, that is, the linear approximation processing means 21, the linear approximation processing data table 22, the diagnostic plant estimated value storage means 23, the linear approximation processing parameter storage means 24, and the plant data change trend Investigation data table 25, discontinuous point detection means 26,
The random point detecting means 27, the sudden change detecting means 28, the gentle change detecting means 29, and the sudden change detecting means 30 can be configured as a separate computer system from the other parts.

Furthermore, in the above-described embodiment, when a change trend determination message indicating "unanalyzable" is obtained, the external procedure interface 92 notifies the operator to that effect, and similarly to the conventional apparatus, the external procedure interface 92 issues data for change trend determination. It is also possible to perform recording output.

[Effects of the Invention] As described above, according to the present invention, a means for determining a change tendency of plant data is provided, and the determination result is used to perform diagnostic processing in a rotating body abnormality diagnosis expert system. Therefore, there is no need for operator intervention.
Therefore, the burden on the operator is reduced, and
It is possible to suppress variations in the diagnosis results of the rotating body abnormality diagnosis expert system. Furthermore, even if an abnormality occurs that needs to be dealt with quickly, the operator can work with leeway, resulting in the effect of improving the stability of plant operation.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing a plant diagnosis device according to an embodiment of the present invention, FIG. 2 is a schematic diagram illustrating the storage contents of a diagnostic plant estimated value storage means, and FIG. 3 is a linear approximation processing parameter storage means. Schematic diagram illustrating the memory contents of 4th
The figure is a schematic diagram illustrating the contents of a data table for investigating plant data change trends, FIG. 7 is a graph diagram for explaining the processing contents of the random point detection means, FIG. 8 is a graph diagram for explaining the processing contents of the sudden change detection means, and FIG. 9 is a graph diagram for explaining the processing contents of the random point detection means. Fig. 10 is a graph diagram for explaining the processing contents of the sudden change detection means, Fig. 11 is a block diagram illustrating a conventional device, and Fig. 12 is a graph diagram for explaining the processing contents of the sudden change detection means. FIG. 13 is a schematic diagram illustrating the contents of the data storage means for data storage, FIG. 13 is a schematic diagram illustrating the contents of the anomaly detection determination data table, FIG. 14 is a schematic diagram illustrating the contents of the anomaly detection data name table, and FIG. Figure 1 is a schematic diagram illustrating the contents of the data table for collecting plant data around the abnormality diagnosis time.
FIG. 6 is a schematic diagram illustrating the storage contents of the diagnostic plant data storage means. 8... Means for collecting plant data around abnormality detection time, 9
...Expert system, 10...Diagnostic plan 1-data storage means, 2l...Line approximation processing means, 2
2... Data table for linear approximation processing, 23... Plant estimated value storage means for diagnosis, 24... Linear approximation processing parameter storage means, 25... Data table for investigating plant data change trends, 26...・Discontinuous point detection means, 27... Random point detection means, 28... Sudden change detection means,
29... Gentle change detection means, 30... Sudden change detection means. Funeral map Funeral area 4 Figure 6 Figure 12

Claims (1)

[Claims] A rotating body abnormality diagnosis expert system that detects abnormalities occurring in the plant based on plant data input from the plant, and queries the results of the detection and the change trends in the plant data within a predetermined range from the time the abnormality was detected. In a plant diagnosis device for diagnosing the cause of an abnormality based on the abnormality detection time, the plant data collection means around the abnormality detection time collects plant data in a predetermined range before and after the time when the abnormality is detected, and the plant data collection means around the abnormality detection time collects plant data around the abnormality detection time. a diagnostic plant data storage means for storing diagnostic plant data; a linear approximation processing means for calculating an estimated value by linear approximation for determining a plant data change tendency based on the stored contents of the diagnostic plant data storage means; A diagnostic plant estimated value storage means for storing estimated values calculated by the linear approximation processing means; a linear approximation processing parameter storage means for storing the linear approximation processing parameters calculated by the linear approximation processing means; a discontinuous point detection means for detecting the occurrence of a discontinuous point in the plant data based on the stored contents of the plant data storage means for use and the plant estimated value storage means for diagnosis; is activated when a point is being detected, and detects the occurrence of a random point in the plant data based on the stored contents of the diagnostic plant data storage means and the diagnostic plant estimated value storage means, and also detects the random point. Then, random point detection means notifies the rotating body abnormality diagnosis expert system of the fact;
This random point detection means is activated when no random point is detected, and detects that a sudden change has occurred in the plant data based on the stored contents of the diagnostic plant estimated value storage means and the linear approximation processing parameter storage means. and sudden change detection means for notifying the rotating body abnormality diagnosis expert system when a sudden change is detected; and said straight line approximation processing parameter storage means that is activated when said discontinuous point detecting means does not detect a discontinuous point. a gentle change detection means for detecting that a gentle change has occurred in the plant data based on the memory contents of the plant data, and notifying the rotating body abnormality diagnosis expert system when a gentle change is detected; is activated when the change detection means does not detect a gentle change, and detects that a sudden change has occurred in the plant data based on the stored contents of the linear approximation processing parameter storage means, and when a sudden change is detected, the A plant diagnosis device characterized by comprising a sudden change detection means for notifying a rotating body abnormality diagnosis expert system.