JPH0293709A - Supporting device for process plant operation - Google Patents

Abstract

PURPOSE:To diagnose even concerning unknown abnormality to take countermeasure against the abnormality by providing a plant data base, a knowledge base to house the function hierarchical structure knowledge of a plant as well, an inferring mechanism and a man machine interface. CONSTITUTION:In a knowledge base 4, a knowledge 7 to describe the function hierarchical structure of the process plant is housed in addition to knowledges (figures 8-12) to be needed for the abnormality diagnosis and countermeasure plan of the process plant. For this knowledge 7, a nuclear power plant 1 is grasped as the hierarchical structure from the side of the function according to the view point of knowledge engineering. Thus, in an inferring mechanism 5, on the basis of this knowledge 7, the retrieval and diagnosis of the plant abnormality, the selection of an operation target and further, the plan and synthesization of the countermeasure are systematically executed from the view point of production and security to be the original purpose of the plant. Thus, interaction between man-machine is easily executed and the enough processing can be executed even concerning the unknown abnormality which is not forecasted in advance. Then, the optimum countermeasure can be presented.

Description

DETAILED DESCRIPTION OF THE INVENTION (Objective of the Invention) (Industrial Application Field) This invention is directed to the production of substances or energy using physical or chemical processes, such as power plants and oil refinery plants. During the operation of the bran 1- that produces the product, process disturbances caused by equipment failures, malfunctions, incorrect operation of the operating port, disturbances, etc. are detected, the cause of the abnormality and the state of the bran 1- are diagnosed, and these This invention relates to a process plant operation support 4I device that selects a new operation target from the results of and formulates countermeasures necessary to achieve the target.

(Prior art) If the disturbance of the prohis plant is left untreated, it will cause imbalance of substances and concentration of energy in local parts of the plant, which may damage the plant or release harmful substances 71 into the environment. It is necessary to detect this disturbance at an early stage, diagnose the state of the plant, and take appropriate measures.

Therefore, in conventional process plants, abnormalities are detected and diagnosed by collecting necessary information within the plant in a central control room and monitoring this information. In other words, a normal operating range is determined in advance for measured values related to necessary information from the plant, and a ¥1 alarm device is installed that issues a warning to operators when the measured value deviates from this normal operating range. I am able to monitor it. Furthermore, methods for formulating countermeasures in the event of an alarm are dealt with through procedures such as written procedures or driver training.

However, in recent years, as plants have become larger and more complex, the scale of the above-mentioned alarm systems has also increased, making it even more difficult to handle! ! Books and training too? J2 is becoming more complex. Therefore, when a disturbance occurs, a large number of alarms are generated that far exceed the processing capacity of the operating port, making it difficult for operators to respond. In response to this trend, systems are being developed that utilize computer information processing technology to provide only truly valuable and appropriate information to operators. However, this is considered to be only a partial response overall. So, for example, what has been proposed as an alarm suppression system:
Select the earliest alarm from a large number of alarms, or select unnecessary alarms based on relatively small logic (for example, when a "bulky" alarm occurs, a "high" alarm of the same signal is deleted). ! ! Although it is acknowledged to be effective, it cannot be considered as a drastic countermeasure.

On the other hand, in general, an abnormality diagnosis system using a knowledge engineering method uses a knowledge base containing knowledge about the causes of abnormalities and abnormal changes that appear in plant measurement signals as a result, and 11. The diagnosis is made by For example, when an abnormal pattern is obtained from the measured values of a plant, the inference mechanism searches the knowledge base to determine whether there is knowledge that matches that pattern, and picks up the one that matches. If there are multiple pieces of matched knowledge, the most probable one is selected to estimate the cause of the abnormality.

(The problem that FE Ming is trying to solve) However, with such an abnormality diagnosis system,
It is possible to diagnose abnormal causes that have been considered in advance and stored in the knowledge base, but it is not possible to diagnose causes that have not been considered in advance. Furthermore, the amount of knowledge regarding the causes of abnormalities and patterns of symptoms increases dramatically as the scale of the system increases, resulting in an enormous amount of processing time.

This invention was made in consideration of the above-mentioned circumstances, and allows for the detection and diagnosis of plant abnormalities, the selection of operational targets, and the formulation of countermeasures over a wide range of areas and quickly. It is an object of the present invention to provide a process plan i/driving support device that can also diagnose causes and abnormal patterns and deal with the abnormalities.

[Structure of the invention]

(Means for Solving the Problems) This invention provides a plant database that stores plant data from a process plant, and the process plan 1 described above.
A knowledge base equipped with the knowledge necessary for diagnosing abnormalities and formulating countermeasures for ~, an inference mechanism that uses the above plant database and knowledge base to diagnose abnormalities and formulating countermeasures based on knowledge engineering methods, and the results of the Jff theory. It has a man-machine interface that allows operators to communicate with the operating port, and the knowledge base includes the functions of the process plant! It is characterized by storing knowledge describing the IJ layer structure.

(Function) Therefore, according to the driving support device for Pro Wrestling Plan 1 according to the present invention, abnormality diagnosis using the knowledge engineering method,
By systematically carrying out production activities, which are the original purpose of the plant, and ensuring safety, it is possible to broadly and quickly grasp the plant status, including unknown conditions that have not been considered in advance, and to meet operational targets. It is possible to select the desired plant state, plan and evaluate the necessary measures to be taken to achieve the goal, and provide this information to operators via an appropriate man-machine interface.

(Example) Embodiments of the present invention will be described below based on the drawings.

FIG. 1 is a block diagram showing a nuclear power plant operation support system which is an embodiment of the process plant operation support system according to the present invention applied to a nuclear power plant.

The nuclear power plant operation support device 2 includes a plant database 3 that stores plant data from the nuclear power plant 1 as a process plant, and a knowledge base that has knowledge for diagnosing abnormalities and formulating countermeasures for nuclear power generation plans 1 to 1. 4 and the above plant database 3
and an inference mechanism 5 that uses the knowledge base 4 to diagnose abnormalities and formulate countermeasures, and a man-machine interface 6 that provides the results of the inference to the operator.

Since the nuclear power plant 1 is a system that generates electricity using nuclear fission reactions, it naturally includes a system that maintains the chain nuclear fission reaction, a system that converts the nuclear fission reaction into thermal energy and transports this thermal energy, and a system that converts the nuclear fission reaction into thermal energy and transports this thermal energy. systems that convert mechanical energy into mechanical energy (rotation of a turbine generator), and systems that convert mechanical energy into electrical energy. Furthermore, each of these systems consists of subsystems that realize subgoals to achieve each goal, subsystems that are further divided into these subsystems, or auxiliary systems that maintain the functions of each of these subsystems. Ru. The current operating conditions of these systems are then input into the plant database 3 as process measurement value signals at appropriate time steps.

As shown in Fig. 4, the knowledge base 4 includes knowledge (hereinafter referred to as hierarchical structure knowledge) 7 that describes the functional hierarchical structure of the nuclear power generation plan 1-1 that corresponds to multiple operational targets, and these functional hierarchies. Knowledge for diagnosing abnormalities at each node of the structure 7 (hereinafter referred to as node diagnosis knowledge) 8, knowledge for diagnosing abnormalities in the entire plant (hereinafter referred to as overall diagnosis knowledge), and new operation based on the abnormality diagnosis combination. Knowledge for selecting a goal (hereinafter referred to as driving goal selection knowledge) >10, knowledge for formulating countermeasures necessary to achieve the driving goal (hereinafter referred to as countermeasure planning knowledge) 11,
Knowledge (hereinafter referred to as countermeasure evaluation knowledge) 12 for confirming whether the impact of implementing countermeasures matches the operational goal is stored.

Among these, hierarchical structure knowledge is based on the knowledge engineering concept.
The nuclear power plant 1 is understood as a hierarchical structure in terms of its functions. In this example, when converting the functional hierarchical structure of the nuclear power plant 1 into a knowledge base, we will focus on the movement of materials within the plant, which is the basis of a process plant, and the movement of energy through the movement of these materials. As a result, knowledge of a clear functional hierarchical structure is created, as shown in Figure 2. This second figure is
This is an example of hierarchical structure knowledge whose operation goal is electric power production in a steady state of operation. In addition, in this Figure 2, flow functions such as transportation, storage, branching, and barriers are shown.
FunCLiO) is expressed using the symbols shown in FIG.

In Figure 2, the functional hierarchical structure for electricity production is as follows:
Flow 5structure
) multiple nodes with specific layer planes 13゜14
．． 15, 16. 17, 18. ..., and these multiple nodes are configured in a hierarchical structure. Each node (flow structure p) 13-18. As the internal structure of...
The flow function shown in Figure 3 and these 70-
centre? support for maintaining the functionality of the function.

For example, the top power production flow 1-Rakti 1213
has a function for achieving power production, and a power production management flow structure 1114 is located at the bottom of this power production 70-structure 13. The power production flow structure 13 includes a source 19 that generates energy through a nuclear fission reaction, and a system 20 that converts and transports that energy.
and has.

The power production management flow structure 14 also includes a nuclear energy generation system 21, a thermal energy transport system 22, an energy conversion system 23, and the like, and each of the systems 21 to 23. ..., respectively, as an auxiliary system necessary to maintain its functions H24.2
5,26. ... will be provided. Additionally, these supports also have flow structures underneath them for implementing each single node. For example, the support 24 has flow structures 15 to 18 . ···have. And this flow structure t15-18
゜... further has a flow function and support inside it. For example, loin structure 15 is
It has flow functions 27, 28, . . . inside it. In this way, a hierarchical structure is constructed.

Now, although the nuclear power plant operation support device 2 has an inference mechanism 5, a flowchart showing an outline of the processing here is not shown in FIG. First, while referring to the plant database 3, the inference mechanism 5 sequentially descends the hierarchy starting from the top of the hierarchy from Plan 1 based on the hierarchical structure jΔ knowledge 7 of the knowledge base 4, and calculates the flow function and the hierarchy in a top-down manner. Determine support health (
Step 1). This reasoning process is similar to the operator's decision-making process. This determination result is temporarily stored in a working memory within the plant database 3, for example.

Then, the inference m4'li5 uses the node diagnostic knowledge 8 based on the health determination result to calculate the flow 1-rachi1=13.14 that includes the flow function or support determined to be unhealthy. Perform abnormality diagnosis for ... (Step 2). 70-S1-Rough Char with Abnormality 13.14. After diagnosing . . . , based on the diagnosis results, an abnormality diagnosis for the whole brand is performed according to the overall diagnostic knowledge 9 (step 3). after that,
Based on the abnormality diagnosis result, the +tt theory mechanism 5 selects a new driving target according to the driving target selection knowledge 10 (step 4), and after implementing the countermeasure, formulates a countermeasure for the driving target according to a11 (step 5).

Next, the inference mechanism 5 determines whether the impact of implementing the countermeasure designed as described above (the plant state after implementing the countermeasure) matches the selected operation target. Evaluate. For that purpose, first, the floor FIvi
Based on the information such as mass and energy on i-Building Knowledge 7,
Qualitatively simulate the planned countermeasures (step 6). And the result of this simulation is
It is evaluated based on the countermeasure evaluation knowledge 12 whether or not the selected driving target is met (step 7). Here, if it is evaluated as matching, J
The driving target and countermeasures before the simulation are completed and the inference result of the inference mechanism 5 is completed. Also,
If it is determined that they do not match, a new countermeasure is drawn up using the countermeasure planning knowledge 11 again for the new plant state obtained as a result of the simulation. In order to obtain the results of this new countermeasure, qualitative simulation is performed again based on the knowledge 7 up to the hierarchical layer, and it is evaluated whether the results match the driving target.

- Repeat these steps (step 5, step 6, step 7) until a.

However, since the results of the qualitative simulation do not always exactly match the operation target, it is necessary to configure the countermeasure evaluation knowledge 12 so that there is a range in the values of plant parameters when performing the evaluation. Additionally, the processes from step 1 to step 7 above are executed at appropriate time steps, but if the simulation results and the driving target are repeatedly evaluated (step 7), if the two do not match forever. In this case, the process is terminated at a reasonable step j, and the driving target at that time and the highest-ranking countermeasure corresponding to the target are selected.

Finally, the inference results of the nuclear power plant blunt operation support system 2 are appropriately displayed to the operator using the man-machine interface 6. Examples of displayed information include the location of the abnormal node in a3 on the plant's functional hierarchy structure and its connections, the abnormal flow function in a3 that covers the existence of the abnormal node, and the abnormal flow function in a3. The location of the support, the connection to these abnormal Zabo 1s, that is, the causal relationship and the range affected by the abnormality, and the new driving goals selected based on these diagnostic results and the measures to be taken to achieve the driving goals. countermeasures to be taken, their purpose and reasons for deriving them, etc.

According to the above embodiment, the nuclear power generation plant l-operation support system N2 uses the knowledge engineering method, that is, the knowledge base 4 and the reasoning mechanism 5, and stores the knowledge 7 of the functional hierarchical structure of the process plant in the knowledge base 4. , plan based on this hierarchical structure knowledge 7!・In order to systematically search for and diagnose plant abnormalities, select operational targets, and formulate and synthesize countermeasures from the perspective of production and safety, which is the original purpose of can. At the same time, compared to conventional methods that use knowledge of individual causes and effects of anomalies, there is no need to assume the causes of anomalies in a single ship, and therefore unknown causes of anomalies can be adequately dealt with.

In addition, since plant abnormalities are diagnosed using a plant model (hierarchical structure knowledge 7) that hierarchically represents the functional structure of the nuclear power plant 1, the diagnosis is performed in order of importance and without exception. It can be executed without any

Furthermore, since the hierarchical structure knowledge 7 is created based on quality a and energy balance, which are essential and proverbial for the nuclear power plant 1, although it is at a high level of abstraction, the structure of the hierarchical structure knowledge 7 is It is characterized by being unambiguous and easy to understand.

In addition, the recommendation process in the recommendation 811 structure 5 of the nuclear power plant operation support system 2 is in line with the decision-making process of operation q, and the intention of the operator is reflected in the reasoning process of diagnosis # & position via the man-machine interface. This allows for easy interaction between humans and machines. For example, the operation target can be easily changed by shutting down the plant from output operation II and moving toward cold shutdown.

Furthermore, J3 in the inference machine 1ili5 not only provides countermeasures based on the abnormality diagnosis results (Step 5), but also predicts the impact of countermeasures as humans naturally think (Step 6), and uses the results as driving targets. Since countermeasures are formulated after evaluating whether or not the above is reached (step 7), it is possible to provide optimal countermeasures.

〔Effect of the invention〕

As described above, the process plant operation support device according to the present invention includes a plant database that stores plant data from process plans 1 to 1, and knowledge that includes knowledge necessary for diagnosing abnormalities in the process plant and formulating countermeasures. an inference mechanism that diagnoses abnormalities and formulates countermeasures based on knowledge engineering methods using the plant database and knowledge base, and a man-machine that realizes dialogue with the operation port by providing the results of inference to the operator. Since the knowledge base contains knowledge describing the functional hierarchical structure of the process plant, it is possible to detect and diagnose plant abnormalities, select operational targets, and formulate countermeasures in a wide range of areas and quickly. Furthermore, it is possible to diagnose unknown abnormality causes and abnormality patterns that have not been predicted in advance, and to adequately deal with the abnormality.

A diagram showing an example of hierarchical 4M-built knowledge in a knowledge base that is the operation target, Figure 3 is a diagram showing symbols of the flow structure used in Figure 2, and Figure 4 is a flowchart showing the configuration of the knowledge base and inference 14M. FIG.

1... Nuclear power plant, 2... Nuclear power generation blunt operation support device, 3... Plant database, 4
...Knowledge base, 5. Reasoning mechanism, 6. Man-machine interface, 7. Hierarchical structure knowledge of nuclear power generation plan 1~, 12. Countermeasure evaluation knowledge.

Applicant's agent Hisashi Hatano

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the installation of a nuclear power plant operation support 11, which is an embodiment in which the process plant operation support device according to the present invention is applied to a nuclear power generation plan i~;
The figure shows electricity production under steady-state operation.

Claims (1)

[Claims] A plant database that stores plant data from process plants, a knowledge base that has the knowledge necessary for diagnosing abnormalities in the process plant and formulating countermeasures, and diagnosing abnormalities based on knowledge engineering methods using the plant database and knowledge base. and an inference mechanism for formulating countermeasures, and a man-machine interface for realizing dialogue with operators, such as providing the results of inference to operators. A process plant operation support device characterized by storing written knowledge.