JPH02224096A - Operation supporting system in abnormal time of plant - Google Patents

Abstract

PURPOSE:To support the condition judgement of an operator by executing only a diagnostic means when a warning is not generated in the case that abnormality is detected, executing both the diagnostic means and a warning processing means when the warning is generated and displaying both results of these means. CONSTITUTION:When the warning is generated, as the output of a warning processing diagnostic part 4, a first generation warning and most important warning, which inquires most attention to the operator, are obtained. When data exist to be coincident with a reference pattern in a reference pattern data base 11, a cause why the first generation warning is generated or additional information to be obtained when the operator experienced such a condition in the past, for example, the name of the warning narrowed down out of the candidates of plural first generation warnings, etc., are obtained as referring data. When the warning is generated, a guide selection part 5 selects a countermeasure to the most important warning out of an operation guide data base 9 and when the event of the abnormality cause is cleared, an operation guide is selected to the event of which a number is designated in the reference pattern data base 11. Besides the operation guide, the detailed diagnostic result and a diagnostic reason are displayed by an output display part 6. Thus, the condition judgement of the operator can be supported.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to an operation support system when an abnormality occurs in a large-scale plant such as a nuclear power plant or a thermal power generation Bran 1-9 chemical plant.

(Prior Art) When a sudden abnormality occurs in a large-scale plant6, its influence is observed as transient changes in multiple signals. If there are many cases of 2 and 2, and the change is large, multiple alarms will be issued. Plant operators are required to quickly judge the status of the plant from multiple alarms such as this and t19, and take appropriate responses. If this is large, the load placed on the operator will be excessive and may lead to erroneous judgments. For this reason, an alarm processing system is being developed for the purpose of assisting operators in making decisions in the event of an abnormality.

(Problem to be Solved by the Invention) Among the above-mentioned conventional warning systems, systems have been developed that determine the first alarm and the latest alarm, and predict future alarms, but these are manuals for handling important alarms. display, and did not have a function to determine the triggering event of the first alarm. For this reason.

Unless the first alarm itself indicates the occurrence of a specific abnormality, operators should use their knowledge of the plant and their own experiential knowledge to identify abnormalities at the equipment/detector level based on the subsequent course of events, etc. The cause had to be determined, leaving the possibility of misjudgment. In addition, conventional alarm systems have been developed in which the order of alarm occurrence for each abnormal event assumed in advance is stored as a database, and the cause of the abnormality is determined by comparing the sequence of observed alarms with the database. . Such systems have the advantage of being able to perform high-speed processing when unexpected events occur, but it is difficult to anticipate all events and cannot respond when unexpected events occur. However, there was a problem in that the cost of creating and managing the database was high.

On the other hand, if a small-scale abnormality occurs in a plant or if it gradually develops, if signs of the abnormality that appear in the process site can be detected early and appropriate measures can be taken, unnecessary plans 1 to 1 can be avoided. Due to the possibility of avoiding suspension,
Early abnormality diagnosis systems have also been developed.

However, in order to display and provide appropriate countermeasures to plant operators in the event of any abnormality,
There was a demand for a system that integrated the diagnostic functions of the conventional early abnormality diagnosis system and alarm processing system.

The present invention was made in view of the above circumstances, and its purpose is to detect an abnormality at an early stage when an abnormality occurs in a large-scale plant, determine the cause, and display necessary countermeasures. When an alarm occurs, it determines the first alarm, predicts future alarms, selects important alarms that require treatment, and simultaneously determines the event that caused the first alarm.
[Structure of the invention] (Means for solving the problem) The above object is to provide an abnormal operation support system for a large-scale plant that supports operators' judgment of the situation by displaying these results. In order to achieve this, the plant abnormal operation support system of the present invention inputs various analog/digital signals representing plant alarm signals, 9 status signals representing operation/stopping of equipment, and 2 representing process quantities.

A signal monitoring means for detecting abnormal changes that appear in these signals and a comparison between the pattern of the signal group in which an abnormality is detected and a reference pattern given in advance for various abnormality-causing events. We use diagnostic means to diagnose the cause of an abnormality at the equipment/detector level, and knowledge about the cause, result, and treatment given in advance for each alarm to identify the propagation route of the alarm that has occurred. an alarm processing diagnostic means that determines the first alarm, predicts future alarms, and selects important alarms that require treatment; A registration means for registering as a pattern is provided, and when an abnormality is detected by signal monitoring, if no alarm is generated, only the diagnosis means is used, and if an alarm is generated, the diagnosis means and the alarm processing diagnosis are carried out. The present invention is characterized in that the means are executed together and the results are displayed together on a display device to support the operator's judgment of the situation.

(Function) According to the plant abnormal operation support system of the present invention,
When an abnormality occurs in a large-scale plant, the system detects the abnormality at an early stage, determines the cause, and displays the necessary measures, allowing operators to take appropriate measures early and avoid unnecessary plant shutdowns. It is expected that there will be more opportunities to avoid this and that the operating rate will improve. In addition, if multiple alarms occur, the most important alarm will be selected and the necessary action will be displayed.
- Not only alarms, but also their triggering events and information about past experiences are displayed as reference data;
It is expected that this will increase the reliability of the operator's judgment of the situation and reduce the possibility of erroneous operation.

(Example) The present invention will be explained in detail by examples.

FIG. 1 is a functional configuration diagram of an embodiment of the present invention.

The plant abnormality support system of the present embodiment is activated at regular time intervals, and three types of plant signals are input from the plant 1 to the signal input section 2: an alarm signal 1, an equipment status a signal, and a process signal. The signal monitoring unit 3 uses two abnormality detection means to check whether an abnormality has occurred in the plant based on attribute data such as the number, type, and abnormality judgment conditions in the input device of each signal stored in the signal attribute database 7.

- After an abnormality is detected, the abnormal signal pattern is input to the signal pattern determination section 10 that performs diagnosis by the diagnostic means, and when an alarm is generated, the alarm is input to the alarm processing diagnosis section 4. Create a pattern. The alarm pattern indicated the values of the alarm signal and the equipment status signal (ONloFF
is expressed by 110), and the abnormal signal pattern is the data with the same content as the alarm pattern, the process signal abnormality judgment result (normal/abnormal expressed by 110), and the object of confirmation in the signal pattern judgment unit 10. It consists of the number of abnormal events that should be considered. This confirmation target event number is used by registering in the signal attribute database 7 as an event number the reference pattern of which abnormal event each signal is used for, and checking all abnormal signals (alarm signals that are ON and It is determined as the union of the event numbers of the process signals determined to be .

The signal pattern determination unit 10 executes the diagnostic procedure as follows. Now, assume that one of the events to be confirmed is event A. The reference pattern data includes the reference pattern of event A, that is, the signal number Si (i:=1-Na) necessary to confirm whether event A is the true cause, and its reference value Ri in the signal pattern. The cause determination threshold TA, which will be given later,
When the output display manual number is registered and the value of Sj in the current signal pattern is assumed to be Xi, the evaluation function JA is calculated using equation (1).

If JA is smaller than the cause determination threshold TA, it is determined that the event A is the cause of the abnormality. Note that it is also possible to use distance as an evaluation function in the above pattern recognition (
Merrill, IEEE Trans, Voll, R.
-22 [see 4 Co., 1973].

On the other hand, when the signal monitoring unit 3 detects the occurrence of an alarm, the alarm processing diagnosis unit 4 is activated in addition to the signal pattern determination unit 1O, and the cause of the alarm signal stored in the alarm related knowledge database 8 is activated. The causal relationship between the generated alarms in the alarm pattern is determined based on the knowledge regarding the generated results. That is, here, the cause determination focusing only on the alarm signal is executed using the device status signal as a determination condition, and the first generated alarm and the newest most downstream alarm are determined. In addition, by registering in the knowledge database 8 the importance determined according to the degree of influence that the firing of each alarm has on the operation of the plant,
Among the generated alarms, the alarm to which the operator should pay the most attention is selected. Here are two alarms: 1. The causal relationship between A2 includes, for example, if the occurrence of the alarm AI is a normal sequence of the safety protection system or interlock mechanism, such as ``If a pump trip J (AI) occurs, a reactor scram J (A2) will occur.'' In the case where the operation causes the occurrence of warning A2 with a 100% probability, and in the case where the r tank liquid level J (A
There are two types of cases where the two are connected by a physical causal relationship, such that even if I) occurs, [tank liquid level J (A2) at most] does not necessarily occur. If the alarm A2 is not generated even after a predetermined delay time has elapsed after the alarm A1 is generated, it can be determined that the sequence is abnormal. Therefore, by storing in the knowledge database 8 the type of causal relationship between alarms, including the delay time if it is due to a sequence operation, it becomes possible to determine a sequence abnormality. In the processing section, when the occurrence of the alarm A1 is confirmed, the presence or absence of an alarm ^2 that would be generated by a sequence operation is checked, and if the alarm I4A2 is present, it is determined as a predicted alarm, and its delay time is registered. From the next time onwards, the state monitoring section counts down the delay time every time a signal is input, and checks whether the predictive alarm A2 has occurred after a predetermined period of time has elapsed.

If a request is made by an operator when the signal pattern determination unit 10 does not determine a cause event, the information obtained through the above alarm processing and the generated alarm pattern will be determined only if there is no abnormality in the sequence operation at that time. A feed pack phenomenon between process quantities that is registered as a reference pattern with additional information provided by the operator.

Alternatively, if there is a closed loop in the cause and effect relationship of alarms due to interlock sequence operation, it may not be possible to identify the first alarm, but it is possible to identify the first alarm that has been narrowed down by the operator's judgment from the results of alarm processing. The above additional information can be registered. This information is displayed as reference data when the same alarm pattern occurs later.

As is clear from the above, when an alarm occurs, the first occurrence alarm and the most important alarm that the operator should pay attention to are obtained as the output of the alarm processing diagnosis section 4, and the reference patterns in the reference pattern database II are determined. = If there is a match, the cause of why the first alarm occurred, or additional information from past experiences, such as the name of the alarm narrowed down from multiple candidates for the first alarm. , will be obtained as reference data.

When an alarm occurs, the guide selection unit 5 selects a treatment for the most important alarm from the operation guide database 9.
If the event causing the abnormality is found, the operation guide for the event whose number is specified in the reference pattern database 11 is selected.

In addition to the selected operation guide, the output display section 6 displays detailed diagnosis results and reasons for diagnosis at the operator's request.

The operation of the alarm processing diagnostic section 4 will be explained with reference to the processing flow diagram of FIG.

The plant status is periodically taken in together with the alarm diagnosis start command 20 and compared with the predicted alarm data 28 to determine whether [P measurement alarm exists or not] 21 3 If it is determined that the predicted alarm exists, it is further compared with the alarm pattern 29 3. If there is no predictive warning, [
"Set display data for sequence abnormality" 23. After the above-mentioned predictive warning processing, the "presence or absence of a new alarm occurrence" 24 is determined.7 If it is determined that a new alarm has occurred, it is used in the alarm-related knowledge database 8 to display the "first occurrence alarm, After carrying out ``Judgment of generated alarm and most important alarm'' 25, further execute ``T-measurement of next generated alarm'' 26, and the alarm diagnosis ends 27.
do.

〔Effect of the invention〕

As explained above, according to the plant abnormal operation support system of the present invention, when an abnormality occurs in a large-scale plant, it is possible to detect the abnormality at an early stage, determine the cause, and display necessary measures. This is expected to improve reactor availability, as operators can take appropriate actions early and avoid unnecessary plant shutdowns. is selected and the necessary action is displayed. In addition, not only the first alarm but also its cause event or information on past experiences are displayed as reference data, making the operator's judgment of the situation more reliable. It is possible to increase the accuracy and reduce the possibility of erroneous operation.

[Brief explanation of the drawing]

Fig. 1 is a functional configuration diagram of an embodiment of the present invention, and Fig. 2 is a functional configuration diagram of an embodiment of the present invention.
- It is a process flow diagram explaining the operation|movement of the alarm processing diagnostic part of figure. ],...Bran 1-1 2...Signal input section 3・
- Signal monitoring unit, 4... Alarm processing diagnosis unit 5... Guide selection unit, 6... Output display unit 7... Signal attribute database 8... Alarm related knowledge database 9... Operation guide database 10... Atonement pattern determination section. 11...Standard pattern database (8733) Representative patent attorney Yoshiaki Inomata (and 1 other person) Figure

Claims (1)

[Claims] (1) Signal monitoring means that inputs plant alarm signals, status signals indicating equipment operation/stoppage, and various analog/digital signals indicating process quantities, and detects abnormal changes that appear in these signals; A diagnostic means for diagnosing the cause of an abnormality at the equipment/detector level by comparing and matching the pattern of the detected signal group with a standard pattern given in advance for various abnormality-causing events; Identify the propagation route of the generated alarm using the knowledge about the cause and result, importance level, and treatment, determine the first alarm, predict future alarms, and select important alarms that require action. and a registration means for registering a generated alarm pattern and a diagnosis result by the alarm processing diagnosis means as a reference pattern for the diagnosis means, and an alarm is generated when an abnormality is detected by signal monitoring. If the alarm has not occurred, only the diagnosis means is executed, and if an alarm has occurred, the diagnosis means and the alarm processing diagnosis means are both executed, and these results are displayed together on the display device to help the operator. An abnormality operation support system for large-scale plants such as nuclear power plants, which is characterized by supporting situation judgment.