JPH05149762A - Malfunction diagnostic device - Google Patents

Abstract

PURPOSE:To enable an input disturbance to a simulator to be automatically set by extracting an abnormal sign and an abnormal cause candidate from an amount of processes of a plant and a most possible abnormal cause to be detected by comparing a simulation result by the simulator with an observation value. CONSTITUTION:A device 1 for inputting an amount of processes inputs the amount of processes from a plant 100. An abnormal sign detection device 3 detects an abnormal sign and extracts an abnormal cause candidate with an observation value of the amount of processes as an input and by using knowledges stored in a first memory. An input disturbance setting device 5 uses the knowledge stored in a second memory 4 for setting the input disturbance from the abnormal cause candidate to a high-speed simulator 6. The high-speed simulator 6 executes simulation by the input disturbance and the observation value of the amount of processes. A comparison device 8 compares the simulation result with the observation value, a device 9 for determinating a cause determines the cause according to the above comparison result, and then an output device 10 outputs it.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an abnormality diagnosis device applied to abnormality diagnosis of various plants such as nuclear power plants.

[0002]

2. Description of the Related Art A nuclear power plant is equipped with an automatic safety protection device based on the multiple protection concept and is operated by a well-trained operator, and exhibits extremely high safety and availability. Recently, in addition to this, efforts are being made to further improve safety by supporting abnormality diagnosis using artificial intelligence technology and simulation technology.

As an application of the artificial intelligence technology, there is an attempt to realize it as a so-called expert system that incorporates and uses the expert intelligence of experts in a specialized field. In this expert system, whether physical quantities such as temperature and pressure, various control signals or manipulated variables such as valve opening and pump speed (hereinafter collectively referred to as process quantities) deviate from a predetermined threshold range. Knowledge for inferring the cause of an abnormality is usually used by a logical combination (logical sum, logical product, negation, implication, etc.) regarding the logical value of. Therefore, knowledge for inferring the cause from the behavior of the process amount in the abnormal state must be given in advance. In other words, if a logical combination other than the given ones occurs, the abnormality cannot be diagnosed.
In addition, the threshold value is almost always given as a fixed value such as an alarm set value, and since it is only improved to the extent that it can be statically changed using the plant load as an index at most, it does not cause erroneous diagnosis during transient phenomena. There is a risk that the diagnosis will be delayed because there is no choice but to set a sufficiently large threshold range. Furthermore, since such a diagnosis is performed by processing the process amount observed in each time section, the expert system cannot realize the diagnosis by observing the temporal trend.

As an example of using the simulation technique, there is an example in which an abnormal sign is detected by estimating a normal state of the process amount by a simulator and monitoring a deviation of the process amount from an observed value. However, the simulation technology is a technology that calculates the behavior of the process quantity with this as an input when the plant disturbance including the cause of abnormality is known, and when the process quantity as a result of the plant behavior is given, it is input with that input. It is generally impossible to back-calculate some plant disturbances. These factors may impose a considerable mental load on the plant operator, and may lead to unnecessary plant shutdowns, resulting in a lower operating rate.

[0005]

SUMMARY OF THE INVENTION The present invention makes it extremely difficult to prepare in advance knowledge for every abnormal event that accurately infers the cause from an abnormal sign that has occurred, but causes an abnormal sign. Focusing on the fact that it is relatively easy to enumerate all possible causes, it was made based on the fact that once the cause candidates were known, computer performance was improved by the time it was possible to simulate their effects sufficiently quickly. It is a thing.

Therefore, an object of the present invention is to detect an abnormality sign and extract an abnormality cause candidate and to automatically set an input disturbance to the simulator from this abnormality cause candidate.
Another object of the present invention is to provide an abnormality diagnosis device capable of detecting the most probable cause of abnormality by comparing the simulation result of the simulator with the observed value.

[0007]

An abnormality diagnosing apparatus according to the present invention is a computer-aided abnormality diagnosing apparatus, which includes process amount input means for inputting a process amount from an abnormality diagnosing target plant and knowledge about an abnormality symptom. First to store
Memory, an abnormality symptom detection unit that detects an abnormality symptom from the observed value of the process amount input by the process amount input unit, and extracts an abnormality cause candidate; a second memory that stores knowledge about the abnormality cause; Input disturbance setting means for setting an input disturbance to the simulator for the abnormality cause candidate output from the abnormality sign detecting means, and output from the abnormality sign detecting means for the input disturbance set by the input disturbance setting means A high-speed simulator for simulating an abnormal cause candidate to be generated, a third memory for storing data for comparing a simulation result of this high-speed simulator with an observation value, and a simulation result of the above-mentioned high-speed simulator with an observation value The comparison means and the source that determines the most probable cause based on the comparison results of this comparison means. Determining means, it is characterized in that an output means for outputting the abnormality cause determined by the reason determining means.

[0008]

[Action]

(1) The process amount input means inputs the process amount from the plant and stores the input process amount in the working memory.

(2) The abnormality symptom detection means receives the observed value of the process amount stored in the working memory as an input,
The knowledge stored in the first memory is used to detect an abnormality sign and extract an abnormality cause candidate, and the detected abnormality sign and the extracted abnormality cause candidate are stored in the working memory.

(3) The input disturbance setting means sets the input disturbance to the simulator using the knowledge stored in the second memory, using the abnormality cause candidate stored in the working memory as an input. The input input disturbance is stored in the working memory.

(4) The high-speed simulator receives the input disturbance and process amount observation values stored in the working memory as input, performs high-speed simulation, and stores the result in the working memory.

(5) The comparing means inputs the simulation result received from the high speed simulator and the observation value stored in the working memory, and uses the data stored in the third memory to calculate the simulation result and the observation value. And the comparison result is stored in the working memory. (6) The cause determining means determines the most probable cause by inputting the comparison result stored in the working memory. (7) The output means outputs to the operator the abnormal sign detected by the abnormal sign detecting means and the most probable cause determined by the cause determining means.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a configuration example of an abnormality diagnosis device according to one embodiment of the present invention. The part surrounded by the broken line is the part that is the subject of the present invention.

The abnormality diagnosing device according to the present invention includes a process amount input device 1, a first memory 2 for storing knowledge about an abnormality sign, an abnormality sign detecting device 3, and a second memory 4 for storing knowledge about an abnormality cause. An input disturbance setting device 5, a high speed simulator 6, a third memory 7 for storing comparison data,
It comprises a comparison device 8, a cause determination device 9, an output device 10, a working memory 11, and a data bus 12.

A process quantity input device 1 for inputting a process quantity such as pressure and flow rate is installed in the plant 100 to be diagnosed, and the output of this input device 1 is inputted to the working memory 11 via the data bus 12. The first memory 2 for storing knowledge about abnormality signs is connected to the abnormality sign detection device 3, and this abnormality sign detection device 3 is connected to the working memory 11 via the data bus 12. The second memory 4 for storing the knowledge about the cause of abnormality is connected to the input disturbance setting device 5, and the input disturbance setting device 5 is connected to the data bus 12
To the working memory 11 via.

Third memory 7 for storing the comparison data
Connects to the comparison device 8 and connects the comparison device 8 to the working memory 11 via the data bus 12. Furthermore,
High-speed simulator 6, cause determination device 9 and output device 10
Are each connected to the working memory 11 via a data bus 12.

Next, the operation of the above embodiment will be described with reference to the flow chart shown in FIG. In the figure, the part surrounded by a thin line frame represents element processing, the part surrounded by a broken line frame represents a decision branch, the part surrounded by a thick line frame represents knowledge or data, and
A thin line arrow indicates a processing flow, and a double line indicates knowledge or data reference. (A) When the abnormality diagnosis process is started, the following process is performed at a predetermined time interval until the end command is received (step A1). (B) The time update process of the working memory (WM) 11 is performed (step A2).

(C) The process amount input device 1 is activated to input the process amount from the plant 100, and the input process amount is stored in the working memory (WM) 11 (step A3).

(D) Next, the abnormality symptom detection device 3 is activated, the observed value of the process amount stored in the working memory 11 is used as an input, and the knowledge about the abnormality symptom is used.
The abnormality sign is detected and the abnormality cause candidate is extracted, and the detected abnormality sign and the extracted abnormality cause candidate are stored in the working memory 11 (step A4).

(E) It is judged whether or not there is an abnormality cause candidate (step A5), and if the abnormality cause candidate is extracted, the input disturbance setting device 5 is activated and the abnormality cause candidate stored in the working memory 11 is detected. As an input, the input disturbance to the high-speed simulator 6 is set by using the knowledge about the cause of abnormality, and the set input disturbance is stored in the working memory 11 (step A6).

(F) If no abnormality cause candidate is extracted, the high-speed simulator 6 is reset (not shown in FIG. 2; detailed in section (7)). Processing (step A2). (G) When an abnormality cause candidate is extracted, for all input disturbances stored in the working memory 11,

(H) The high-speed simulator 6 is started, and the abnormality cause candidate is input to a predetermined time interval,
The simulation is performed at high speed and the result is passed to the comparison device 8 (step A7).

(I) Subsequently, the comparison device 8 is started, and the simulation result received from the high-speed simulator 6 and the observation value stored in the working memory 11 are used as input, and the comparison result is used to calculate the simulation result and the observation value. And the comparison result is stored in the working memory 11 (step A8).

(J) It is judged whether or not the comparison with the simulation results regarding all the input disturbances is completed (step A9), and when the comparison is completed, the cause determining device 9 is activated,
The most probable cause is determined by using the comparison result for each input disturbance stored in the working memory 11 as an input (step A10). (K) Finally, the most probable cause is output from the output device 10 to the operator (step A11). Next, details of the function of each unit will be described. [Knowledge about abnormal signs]

The first memory 2 stores knowledge for detecting an abnormality sign from an observed value of the process amount and extracting an abnormality cause candidate. Knowledge of individual signs of anomalies includes:
The following contents are described. (A) Abnormal symptom name A name uniquely assigned to the abnormal symptom. (B) Abnormal mode The abnormal mode name is defined. The abnormal mode is a detailed classification of abnormal signs such as "excess / vibration" or "high / low" indicating the type and degree of abnormality. For each abnormal mode, (i) Judgment method: Method to judge the abnormal mode of this abnormal sign (ii) Candidate causes: Candidates of all the causes that can cause this abnormal mode (iii) Comparison data: For comparison between cause candidates Describe the comparison data name to be used.

An example of the description format of the abnormality sign detection knowledge is shown in FIG.
An example of knowledge described according to this description format is shown in FIG. In the figure, one or more items with + at the end can be described arbitrarily, and each indicates an or relationship.

In the following, following the general terminology of knowledge engineering,
The entire description of FIG. 3 is referred to as a frame, particularly, in this case, an anomaly symptom frame;
Furthermore, the (symptom determination procedure) and the like inside another stage is called a slot value. In the example of FIG. 4, (a) Signs of abnormal water supply flow rate are:

(B) In the abnormal mode of excessive flow rate, the difference between the observed actual flow rate of water supply and the calculated value of water supply determined by the plant load etc. is calculated by executing whether or not the difference is larger than a preset threshold value ε. Signs can be determined, and candidate causes thereof include high feedwater flow controller gain, high feedwater flow monitoring sensor drift, low feedwater flow control sensor, and low steam generator water level sensor. For comparison between possible causes,
Use water supply system comparison data.

(C) In the abnormal mode of flow rate vibration, the symptom can be judged by executing the "water supply flow rate vibration determination procedure" defined outside this frame. High and feedwater flow rate sensor filters have large time constants. Water supply system comparison data is used for comparison between the cause candidates. It means that. [Abnormality symptom detection device 3] Each time the abnormality symptom detection device 3 is activated, (a) all abnormal symptom frames, the symptom described in (b) all determination method facets in the abnormal mode slot The judgment procedure is executed sequentially,

(C) If and only if any one of them holds, then the abnormal symptom name, the abnormal mode, all the cause candidate names in the cause candidate facets and the comparison data names in the comparison data facets are set. , And stores them in the working memory 11 without duplication. [Knowledge about cause of abnormality]

The second memory 4 stores knowledge for setting the input disturbance to the high speed simulator 6 from the abnormality cause candidates extracted by the abnormality symptom detection device 3 described above. The following contents are described in the knowledge about each cause of abnormality. (A) Abnormal cause name A name uniquely assigned to the knowledge related to the abnormal cause. (B) Application target The variable name for applying the input disturbance to the high-speed simulator 6 and the channel number of the connected device. (C) Value setting The procedure name and parameter list for setting the value of the applied disturbance are given. (D) Time setting A parameter for setting a disturbance application time point when performing a simulation, for example, a time required from occurrence of an abnormality to detection of an abnormal sign is given.

FIG. 5 shows an example of the description format of the abnormality cause candidate extraction knowledge.
6 and 7 show examples of knowledge described according to this description format. Hereinafter, the entire description of FIG. 5 will be referred to as an abnormality cause frame. In the example of FIG. 6, (a) when inputting the cause of the abnormality that the gain of the feed water flow controller is high, (b) the variable name FWC-PG is (c) increased by 10 or 20% from the normal value (“+%”).
Value) is applied 3 or 5 minutes before (d) abnormal sign detection. In this case, since there are two set values and two set times, four input disturbances are set by the product. In the example shown in FIG. 7, (a) when an abnormal cause of the sensor drift for monitoring the feed water flow is high, (b) the variable name FWC-PG is (c) gradually increased from the normal value by 1 or 5% (" % L "
Value) is applied from 10 minutes before (d) abnormal sign detection time. [Input Disturbance Setting Device 5] Each time the input disturbance setting device 5 is activated, (a) an abnormality cause candidate is input from the working memory 11, (b) to the corresponding abnormality cause frame, (c) application target The application target name is retrieved from the slot, (d) the value setting procedure name and all parameters are retrieved from the value setting slot, (e) all time setting values are retrieved from the time setting slot,

(F) In the working memory 11, a set of an application target name, a value setting procedure name, an individual parameter and an individual time set value (a plurality of sets can be set corresponding to the product of the number of values and the number of set times). Store. [High-speed simulator 6]

The high-speed simulator 6 has two operation modes, a high-speed simulation mode and a reset mode, which are used properly depending on whether or not the abnormality cause candidate is stored in the working memory 11. (I) In high speed simulation mode,

The input of the abnormality application target name stored in the working memory 11, the set procedure name of the quantitative value and the set of parameters and the set time are input, and the simulation up to the present before the set time is performed. Are stored in the working memory 11. (Ii) Reset Mode The integral element in the simulator 6 is reset so that the output of the simulator 6 matches the observed value of the process amount stored in the working memory 11. [Comparison Data] The third memory 7 stores data for comparing the simulation result and the observed value. The following contents are described in the comparison data. (A) Comparison data name A name uniquely given to the comparison data. (B) Comparative index

The comparison index for comparing the simulation result and the observed value is given by an arbitrary number of combinations of the comparison method, the calculated value name, the observed value name and the index variable. The calculated value name and the observed value name are the name of the calculated value by the simulator and the name given to the observed value of the process amount, which are indicators for comparison. The comparison method is a method of calculating the comparison index from the calculated value. When using the average error area, enter "EA", when using the maximum error, enter "ME", etc. to specify the prepared calculation method. To do. The index variable is a variable that records the calculation result of the comparative index. The average error area is an average error per unit time obtained by dividing the time integration amount of the absolute value of the difference between the calculated value and the observed value by the integration time width. (C) Evaluation index Write a calculation procedure for combining a plurality of comparison indexes into one evaluation index. For example, it is given by the linear combination of the index variable diagram 4 in consideration of the unit conversion of the physical quantity and the evaluation weight. (D) Evaluation criteria

Usually, the determination of the cause is to find the error between the calculated value and the observed value, that is, the smallest one of the evaluation indexes. However, when all the evaluation indexes are very large (for example, due to an unexpected cause, etc.). ), Or conversely, if it is extremely small (either is likely), the comparison becomes meaningless. Therefore, a reference range in which the comparison is effective is provided. FIG. 8 shows an example of the description format of the comparison data, and FIG. 9 shows an example of the data described according to this description format. Hereinafter, the entire description of FIG. 8 is referred to as a comparison data frame. In the example of FIG. 9, (a) the water supply system comparison data is (b) there are two comparison indexes,

Calculated value C_FWF of water supply flow rate and observed value M_
The average error area EA of FWF is calculated and the index variable E_FW is calculated.
Store in F and observe with calculated value C_SGL of steam generator water level
The maximum error ME of the value M_SGL is calculated and the index variable E_S is calculated.
Store in GL. (C) The stored values are stored in the equation a₁× E_FWF + a₂Passed as an evaluation index by × E_SGL. (D) However, the evaluation index (indicated by *) is the evaluation reference range.
(B₁<< b₂The comparison is valid when Is an example of expressing the data. [Comparison Device 8] The comparison device 8 uses the high-speed simulator 6.
Is started every time the simulation is completed once, and (a) all the ratios stored in the working memory 11
The stored data in the third memory 7 is referred to for each comparison data, and (b) the simulation result and the working memory 11 are stored.
With the stored process amount as input, the comparison index slot
(C) Evaluation index calculation given to the evaluation index slot.
Calculate the evaluation index according to the procedure, and (d) set of comparison data name, corresponding input disturbance and evaluation index
Are stored in the working memory 11. [Cause Determining Device 9] The cause determining device 9 is configured to detect all input disturbances.
Is started after the simulation for (a) All the ratios stored in the working memory 11
For the set of comparison data name, corresponding input disturbance and evaluation index, (b) Find the one with the smallest evaluation index for each comparison data.
This is determined to be the cause. (I) The evaluation index is the effective reference score of the comparison data frame.
There are multiple items within the valid reference range given to the lot.
If there are individual pieces, they are all considered to be the cause. (Ii) The smallest evaluation index is the effective basis of the comparison data frame.
If it is over the effective reference range given to the quasi-slot
If the cause is unknown. [Output Device 10] The output device 10 includes (a) an abnormality sign detected by the abnormality sign detection device 3.
And abnormal mode, (b) For each comparison data determined by the cause determination device 9
Outputs the determined cause (including unknown cause). [Working memory 11] Working memory 11 is as follows
Area. (A) Process amount region In the process amount region, (i) observed time: t (ii) multiple observed process amounts: x₁, X₂・ ・ ・
., X_m  Is set as one record, and the number of records is set in advance.
To record. (B) Abnormal symptom detection result area The abnormal symptom detection result area differs from the detected abnormal symptom name.
Stores a set of normal modes. (C) Abnormal cause candidate area In the abnormal cause candidate area, the extracted cause candidate name and comparison data are compared.
Stores a set of data names. (D) Input disturbance area

In the input disturbance area, the name of application target and the value setting
Sequential name, individual parameter and individual time set pairs
(Multiple corresponding to the product of the number of values and the number of set times
Can be assembled. ) Is stored. (E) Simulation result area The input disturbance area is simulated by the high-speed simulator 6.
(I) calculated time step points: t (ii) calculated multiple process quantities: y₁, Y₂・ ・ ・
., Y_n  Is set as one record, and the number of records is set in advance.
To record. (F) Comparison result area In the comparison result area, the comparison data name, the corresponding input disturbance and
Stores a set of evaluation indicators. (G) Determined cause area The determined cause of abnormality is stored in the determined cause area.

[0040]

As described above in detail, according to the present invention, at least the knowledge for accurately inferring the cause from the abnormal sign describes at least the cause candidate that causes the abnormal sign.
It is possible to comprehensively search for causes, find out the most probable cause, and notify the operator, and find not only the situation at each time, but also the optimum solution as seen by experienced operators as well as time trends. it can. Therefore, it is possible to reduce the mental load on the plant operator at the time of an abnormality, to quickly conclude an abnormal event, and to avoid unnecessary plant shutdowns, further improving plant safety and availability. Can be improved.

[Brief description of drawings]

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

FIG. 2 is a flowchart showing the overall processing operation.

FIG. 3 is a diagram showing an example of a description format of knowledge about abnormal signs.

FIG. 4 is a diagram showing an example of knowledge about an abnormal sign described according to the description format of FIG.

FIG. 5 is a diagram showing an example of a description format of knowledge about an abnormality cause.

FIG. 6 is a diagram showing a first example of knowledge about the cause of abnormality described according to the description format of FIG. 5;

FIG. 7 is a diagram showing a second example of knowledge about the cause of abnormality described according to the description format of FIG. 5;

FIG. 8 is a diagram showing an example of a description format of comparison data.

9 is a diagram showing an example of comparison data described according to the description format of FIG.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Process amount input device, 2 ... 1st memory which stores knowledge regarding an abnormal sign, 3 ... Abnormal sign detection device, 4 ... 2nd memory which stores knowledge regarding an abnormal cause, 5 ... Input disturbance setting device, 6 ... High-speed simulator, 7 ... Third memory for storing comparison data, 8 ... Comparison device, 9 ... Cause determining device, 10 ... Output device, 11 ... Working memory, 12
… Data bus, 100… Plant for diagnosis.

Claims (1)

[Claims] 1. An abnormality diagnosing device using a computer, a process amount input means for inputting a process amount from an abnormality diagnosis target plant, a first memory for storing knowledge about an abnormality sign, and the process amount input means. Output from the abnormal sign detecting means, an abnormal sign detecting means for detecting an abnormal sign from the observed value of the process amount input by the CPU, and extracting an abnormal cause candidate, a second memory for storing knowledge about the abnormal cause, and the abnormal sign detecting means. Input disturbance setting means for setting an input disturbance to the simulator for the abnormality cause candidate, and simulation of the abnormality cause candidate output from the abnormality sign detecting means for the input disturbance set by the input disturbance setting means A high-speed simulator and data for comparing the simulation results and observation values of this high-speed simulator A third memory for storing, a comparing means for comparing the simulation result of the high-speed simulator and the observed value, a cause determining means for determining the most probable cause based on the comparison result of the comparing means, and the cause determining means. And an output unit that outputs the cause of the abnormality determined by the abnormality diagnosis apparatus.