JPH01295303A - Operating knowledge collector - Google Patents

Abstract

PURPOSE:To shorten the time required for the interviews between a knowledge engineer and an expert by using a means which produces a knowledge sentence 'state of IF data item/THEN operating item' based on each operation and the state of the data items extracted in relation to those operations. CONSTITUTION:A status sampler 10 retrieves a status data base 5 to take out the corresponding data item based on the sorting of operations received from an operation sampler 7. Then the sampler 10 takes the value of said data out of a data group 1 and gives it to a knowledge generator 9. The generator 9 produces and transmits a knowledge sentence 'state of IF data item/ THEN operating item' based on the operating items, the sorting of operations, and the time points and dates of generation of operation as well as the states of data items received from the sampler 10. Thus it is possible to shorten the time required for the interview between a knowledge engineer and an expert and to quickly form an expert system.

Description

DETAILED DESCRIPTION OF THE INVENTION OBJECTS OF THE INVENTION (Industrial Field of Application) The present invention relates to an operation knowledge collection system suitable for constructing an expert system in process operation.

(Prior Art) The knowledge required for an expert system must fully reflect the empirical rules of experienced experts.

Therefore, in order to collect the knowledge required to construct an expert system for process operation, knowledge engineers must conduct a large number of interviews with experts in the process, and the knowledge collected through the interviews can be used to ``IF THE''
The tedious work of arranging the rules to the NJ style had to be done almost manually without using a mechanical system.

(Problem to be solved by the invention) As described above, in order to collect the knowledge required to construct an exact system for process operation, knowledge engineers and experts in the process had to conduct a huge number of interviews to collect knowledge. As a result, we had to establish a long-term joint development system between knowledge engineers and experts.

Furthermore, when constructing an expert system for a process for which the operation method has not yet been determined, it is extremely difficult to collect knowledge because there are no experts, and as a result, it takes a long time to construct and operate the expert system. There was a problem that it took a long time.

This invention was made to solve the above problems, and its purpose is to reduce the time required for interviews with knowledge engineers and experts in order to collect the knowledge required to construct an expert system for process operation. The purpose of the present invention is to provide a driving knowledge collection device that can speedily construct an expert system by shortening the time and automating the subsequent rule creation work.

[Structure of the Invention] (Means for Solving the Problems) In order to achieve the above object, this invention provides various operations that are essential for plant operation and data items that serve as a standard for determining whether or not to perform the operations. means for registering the data in association with each other; means for displaying the state of the data IQ designated by the registered contents each time each of the registered operations is performed during plant operation; and each of the operations and and means for generating a knowledge sentence of the form "The state of an rIF data item THEN operation" based on the state of the data item extracted in connection therewith.

(Function) According to this structure, in an interview to collect knowledge, the knowledge engineer asks the process expert only what data items are referenced when performing each operation. Since there is no need to ask questions about the specific value of the data item, the time required for the interview can be significantly reduced.

In addition, as long as the various operations required for plant operation are associated with the data items that serve as the basis for determining whether or not to perform the operations and registered, each time each of the registered operations is performed during plant operation, , the state of the data item specified by the registered content is extracted, and a knowledge sentence is automatically generated based on each operation and the state of the data item extracted in relation to it. This eliminates the need to create knowledge sentences at all, frees knowledge engineers from troublesome work, and significantly reduces the time required to create knowledge sentences.

(Embodiment) FIG. 1 is a block diagram schematically showing the configuration of a driving knowledge collection device according to the present invention.

In the figure, data group 1 shows each data taken out from a plant subject to process control via a process input device or the like.

The display input device 2 is composed of a CRT, a keyboard, etc., and is used for registration, modification, and deletion in a status database and an operation database, which will be described later.

Used for display, etc.

The display input device 3 is similarly composed of a CRT, a keyboard, etc., and is used for registration in a knowledge database, RI correction, etc., which will be described later.
Used for deletion 1 display, etc.

The operating device 4 is used to start and stop the equipment, set and change set values, and is comprised of various operating switches, a numeric keypad, and the like.

In the status database 5, data items related to operation items are stored using the operation classification as a key based on the operation of the display input device 2.

The operation database 6 stores operation items and operation classifications (details will be described later) by operations on the display input device 2.

Here, the status database 5 and the operation database 6 are associated with each other using the operation classification as a key. Therefore, as will be described later, each operation item and the data item that serves as the criterion for executing the operation item are They are registered in association with each other.

The operation sampler 7 monitors the execution of any operation item on the operation device 4, and when any operation item is executed, the operation sampler 7 stores data indicating the operation item and operation classification in the operation database. At the same time, the date and time of the operation are taken out from the clock device 8 and given to the knowledge generator 9.

At the same time, the operation sampler 7 assigns the operation classification related to the executed operation item to the status sampler 1.
Give to 0.

In the status sampler 10, based on the operation classification given from the operation sampler 7, the status database 5 is searched to extract the corresponding data item, the value of the data related to the data item is extracted from the data group 1, and this is sent to the knowledge generator. Give to 9.

In Knowledge Generator 9, Operation Sampler 7
Operation item, operation classification, and operation occurrence date given from 2
Based on the time and the state of the data item given from the status sampler 10, [IF data item state J TH
Generates and outputs a knowledge sentence in the format EN r operation item J.

The intelligence database 11 stores knowledge sentences generated and output from the knowledge generator 9, and the stored knowledge sentences can be modified and deleted as appropriate by operating the display input device 3.

Next, we will explain the operation of the driving knowledge collection device with the above configuration.
A systematic explanation will be given with reference to Examples 1 to 5.

Let us now assume that we want to collect operating knowledge regarding the operation of a certain pump that has been incorporated into a process.

In this case, the knowledge engineer asks the expert of the process, for example, "What operation items are there for this pump?"

On the other hand, on the side of experts in the relevant process, for example,
There are two operation items for this pump: start and stop.''

Based on the questions and answers made in this way, the knowledge engineer operates the display input device 2 to
A database as shown in Example 1 can be stored within the operational database 6.

Example 1 The operation classification is pump operation, and the operation items related to pump operation are as follows. 1. The pump was operated. 2. The pump was stopped. Next, the knowledge engineer tells the expert of the process, for example, " Ask questions such as "What data items would you refer to to start or stop this pump?"

On the other hand, experts in the relevant process refer to the pump well water level, predicted pump well water level, inflow flow rate, and predicted inflow flow rate in order to start or stop the pump.''Let's do it.

Based on the questions and answers made in this way, the knowledge engineer can register a database as shown in Example 2 in the status database 5 by operating the display input device W12.

Example 2 The operation classification is pump operation. The data items related to pump operation are as follows: 1, pump well water level 2, pump well water level prediction value 3, inflow basin 4, inflow flow rate prediction value In addition, the above status database 5 ．． The reason why the keys of the status database 5 are classified into operation categories when registering the operation database 6 is to share data items, thereby preventing the status database from becoming too large.

When the registration of the status database 5 and operation database 6 is completed in this way, the knowledge engineer switches the operation mode of the equipment from the registration mode to the knowledge collection mode, and also switches the operation mode of the equipment from the registration mode to the knowledge collection mode.
start operation.

In the knowledge collection mode, when the operating device 4 is operated by the operator to operate the equipment, the operation sampler 7 detects this.

Then, in the operation sampler 7, the operation device 4
The operation items and operation classifications corresponding to the signals from are retrieved from the operation database 6, and in order to facilitate the chronological investigation of knowledge, the operation date and time data are retrieved from the clock device 8, and these are sent to the knowledge generator 9. give.

At the same time, the operation sampler 7 also provides the operation classification degree extracted from the operation database 6 to the status sampler 10.

Then, the status sampler 10 searches the status database 5 for the corresponding data item based on the operation classification 2 given from the operation sampler 7,
Based on the retrieved data items, the status of each data item is extracted from the data group 1 and provided to the knowledge generator 9.

Then, the knowledge generator 9 uses the operation items given from the operation control sampler 7, the operation classification 2 date 1 time, and the data r given from the status sampler 10.
Based on the first state, a knowledge sentence in the form of "IF data item state THEN operation item" as shown in Example 3 is generated and registered in the knowledge database 11.

Example 3 Operation classification Showa Year Month Day Hour 2 minutes: Seconds F (Data item status) THEN (Operation item) What is important here is that the form of the knowledge statement registered at this time is arranged by the knowledge engineer after knowledge collection. The point is that it is a complete thing that does not require.

Further, the knowledge sentences registered in the knowledge database 11 are modified as appropriate by operating the display input device 3.

Deletion etc. are made possible.

Examples 4 and 5 show two examples of knowledge sentences generated in this way regarding pump operation.

Example 4 Operation classification is pump operation IF operation occurred on 1988-2-17 12:30:43 (Pump well water level is 9.81 Pump well water level predicted value is 9.6111 Inflow flow rate is 3.
3i3/S is the predicted inflow flow rate of 3.3m + 3/S) T 116 N (The pump was operated) Example 5 Operation classification is pump operation 1988-2-17 18:38:50 Operation occurred IF (pump The well water level is 9.51.The predicted pump well water level is 9.6 m.The inflow flow rate is 3.21113/S.The predicted inflow flow rate is 3.3 m3/S. ) As is clear from these examples, since the knowledge sentences generated by this device are in the IF THEN format, there is no need for the knowledge engineer to arrange them again, and they can be used as they are for constructing an expert system.

Also, since the date and time of the operation occurrence are entered, data group 1. By comparing the operation items with the external environment situation, which cannot be completely known only from the signals from the operating device 4, it is possible to clearly understand each operation and the situation related to it more accurately at a later date.

In this way, according to the device of this embodiment, when collecting knowledge, the knowledge engineer only needs to ask the experts in the process about the operation classification, operation items, and data items, which reduces the time required for interviews compared to the conventional method. After completing the registration of operation classification, operation items, and data items, the knowledge engineer no longer needs to perform any special operations. This frees up knowledge engineers from tedious work, which in turn makes it possible to build expert systems faster and at lower costs.

[Effects of the Invention] As is clear from the above explanation, according to the present invention, knowledge regarding process operation is automatically collected, the time required for interviews with knowledge engineers and experts can be significantly reduced, and the time required for interviews with knowledge engineers and experts can be significantly reduced. Even in the case of processes that have not yet been determined, it is possible to easily collect knowledge that fully reflects the empirical rules of experts and experts, thereby achieving faster and lower cost construction of expert systems. can do.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of a driving knowledge collection device according to the present invention. 1...Data group 2.3...Display input device 4...Operation device 5...Status database 6...Operation database 7...Operation sampler 8...Clock device 9... Knowledge generator 10...Status sampler 11...Knowledge database

Claims (1)

[Scope of Claims] Means for registering various operations necessary for plant operation in association with data items serving as a criterion for determining whether or not to perform the operations; and each of the registered operations during plant operation. means for extracting the state of the data item specified by the registered contents, and an IF data item state THEN operation based on each of the operations and the state of the data item extracted in connection therewith. A driving knowledge collection device characterized by comprising: means for generating a knowledge sentence of the form ``.''.