JPS63257827A - Problem solving device - Google Patents

Abstract

PURPOSE:To improve remarkably the inference efficiency and memory efficiency by providing the structure of an intelligent data base and a dynamic assignment intelligence utilizing function retrieving only a required minimum intelligent module in the process of solving a problem. CONSTITUTION:An intelligent data base 2 is a set of plural intelligent modules 1 arranging plural intelligences with different property and category to solve a problem as an aggregate for each property and category and has a structure expressing quantitatively the relevance between two optional modules and the relation of the single knowledge belonging to one and other modules in terms of a figure representing the degree of connection. The dynamic assignment intelligence utilizing function 3 extracts or abolishes dynamically only the module required in the process of solving a problem based on the degree of coupling. Input/output device 6, 7 apply input/output of information required for solving the problem.

Description

DETAILED DESCRIPTION OF THE INVENTION Field of Industrial Application The present invention relates to a dossier solving device for solving a specific problem using a plurality of pieces of knowledge having different properties and categories.

BACKGROUND OF THE INVENTION In recent years, knowledge database systems have been studied as an application of artificial knowledge in order to solve specific problems using multiple pieces of knowledge with different properties and categories. In particular, systems for diagnosing failures of process equipment such as semiconductor devices are being put into practical use due to the need for 24-hour operation and rapid failure response by equipment operators other than maintenance personnel.

A conventional configuration will be explained with reference to FIG. FIG. 7 shows the configuration of the knowledge database Nostem for diagnosing equipment failures. The CRT terminal 6 is a device for interactively diagnosing faults, and is used for manually inputting information necessary for fault diagnosis, inputting commands to instruct reasoning methods at the beginning or during the diagnosis process, and inputting commands for instructing reasoning methods and equipment operations. It displays methods, diagnostic results, etc. in text and squares. The central processing unit 8 centrally manages and controls programs for performing failure diagnosis processing for equipment A, the CRT device 6 and equipment 7, and the knowledge database 32. The program that performs fault diagnosis processing includes a knowledge database acquisition function 9 that constructs a knowledge database 32 described in a rule format of 100~then..., and a fault diagnosis function that uses the knowledge database 32 to perform fault diagnosis by inference. Diagnostic function 33
The main configuration is a device communication function 10 that receives information necessary for failure/waterfall diagnosis from the device 7 and transmits information for restoring or controlling the device 7 based on the failure diagnosis result. .

Problems to be Solved by the Invention The advantages of conventional knowledge database systems that perform fault diagnosis are that the individual pieces of knowledge that make up the knowledge database are highly independent, and that the fault diagnosis function, which is the core program that performs fault diagnosis, is highly independent. Since there is no relation to the meta-knowledge that realizes 33, knowledge can be easily updated (added, changed, deleted).

However, since the knowledge is highly independent, inference during fault diagnosis, it is necessary to pattern match all the knowledge with the condition part of the rule, which is inefficient in terms of inference time. Furthermore, it is necessary to make the entire knowledge database resident on the storage device, which poses a problem in terms of memory efficiency.

In addition, as one method to solve the above problems, for example, as shown in Japanese Patent Publication No. 61-26112, the knowledge database is divided into groups from a certain point of view, and the necessary information in the grouped reduced knowledge database is One possible method is to extract only the information in advance, but the amount of knowledge to be matched does not change during the fault diagnosis process, and this does not solve redundancy.

Means for Solving the Problems The present invention overcomes the above-mentioned drawbacks by creating knowledge modules that group a knowledge database from a certain point of view, and also by controlling the degree of connectivity between modules or between single knowledge and modules. The purpose of the present invention is to provide a problem-solving method and device for troubleshooting, etc., which has a knowledge utilization function that dynamically allocates only the knowledge that is truly necessary.

The configuration of the present invention will be explained with reference to FIG. FIG. 1 shows an apparatus embodying the invention. The present invention is a collection of a plurality of knowledge modules 1 in which a plurality of pieces of knowledge with different properties and categories for solving a certain problem are put together as one set for each property and category, and among the plurality of knowledge modules. A knowledge database that has a structure that quantitatively expresses the relationship between any two knowledge modules or the relationship between one knowledge module and a single piece of knowledge belonging to another knowledge module using a numerical value called the degree of connectivity. 2, a knowledge utilization function 3 that dynamically extracts and rejects only the knowledge modules necessary in the process of solving a certain problem based on the degree of connectivity; and a knowledge utilization function 3 that stores the knowledge database 2. a storage device 4 for
'5-A problem solving function 5 and an input/output device 6 for inputting and outputting information necessary for problem solving. It is comprised of a central processing unit 8 that comprehensively manages and controls each of the devices and functions mentioned above.

Function The function of the present invention will be explained with reference to FIGS. 2 to 4. For convenience of explanation, the subject is equipment failure diagnosis. FIG. 2 shows a knowledge database system of the present invention for diagnosing failures of equipment 7. The knowledge database 2 has information such as know-how based on the unique skills and experiences of experts in many fields necessary for fault diagnosis of the device, and the length of the device held and usable within the device 7. The trouble knowledge module 11 is a knowledge module regarding failure items, the global knowledge module 12 is a general and common knowledge module, and the local knowledge module 13 is a plurality of knowledge modules having different properties and categories. Reference numeral 14 denotes a fault diagnosis function having a function 3 for utilizing dynamic side guess knowledge. FIG. 3 shows the contents of the trouble knowledge module 11. The global knowledge module 12 and local knowledge module 13 also have similar contents.

The knowledge module has a degree of connection between other knowledge modules of 15 and knowledge of 16, and a degree of connection between each knowledge and the pond knowledge module of 17.
It is made up of. The value used for the degree of connection is, for example, a value with an absolute value from 0 to 100, and positive and negative are distinguished by a sign. That is, if the degree of coupling is -100, it represents an exclusive relationship. Note that a known method such as statistical probability or Bayesian probability may be used as a calculation method for combining the degree of connectivity. Fourth
The figure shows the basic operation of the invention.

First, knowledge initialization (18) is performed to enable necessary knowledge modules. Next, inference is made within valid knowledge (19). Even if the inference is reached, input of necessary information is requested (2o). In response to this input request, a new fact is fixed by receiving the necessary information manually input by the operator (21). If the fact adopts a certain piece of knowledge, then the knowledge module that combines that piece of knowledge is added to the valid knowledge (2
2).

On the other hand, if that fact rejects (negates) a certain knowledge, the knowledge that derived that knowledge R is also rejected. At this time, the knowledge module to which the rejected knowledge is combined is deleted in favor of valid knowledge (23).

Thereafter, return to step 19 and repeat the above procedure. The termination conditions are
This is a case where no facts are added after inputting the necessary information (20). That is, the process ends when the knowledge for pattern matching no longer exists.

Embodiment An embodiment of the present invention is shown in FIGS. 5 and 6. FIG. 5 shows how knowledge modules are combined. Solid lines indicate positive bonds;
Dashed lines indicate negative bonds. Figure 6 shows the connection 1 shown in Figure 6.
It shows the transition of dynamic allocation of knowledge modules when there is a relationship.

Note that there are various viewpoints when dividing the knowledge database into groups, and it does not depend on the method of grouping. Further, although the present invention has been exemplified as a method for diagnosing equipment failures, it goes without saying that the method can be applied to solving other types of problems.

Effects of the Invention As described above, the present invention improves inference efficiency and memory efficiency by providing a knowledge database structure that searches only the minimum necessary knowledge modules in the process of problem solving and a dynamic side-by-side knowledge utilization function. The practical effects of this technology are extremely large, achieving significant efficiency improvements in terms of efficiency.

[Brief explanation of drawings]

FIG. 1 is a diagram showing a device according to an embodiment of the present invention, FIG. 2 is a diagram showing the configuration of a device failure diagnosis knowledge database system according to the same embodiment, and FIG. 3 is an explanatory diagram of the contents of a trouble knowledge module according to the same embodiment. , FIG. 4 is an explanatory diagram of the procedure on the dynamic side of the knowledge module of the same embodiment, FIG. 5 is an explanatory diagram of the state of combination of knowledge modules of the same embodiment, and FIG. FIG. 7, which is an explanatory diagram of the transition of module dynamic side pressure, is a configuration diagram of a conventional equipment failure diagnosis knowledge database. 1...Knowledge module, 2...Knowledge database, 3...Dynamic side hit knowledge utilization function, 4...Storage device for storing the knowledge database , 6...Problem solving function, 6...CRT terminal, 7...External equipment, 8...Central processing unit. Name of agent: Patent attorney Toshio Nakao and 1 other person01
figure

Claims (1)

[Claims] It is a collection of multiple knowledge modules in which multiple pieces of knowledge with different properties and categories for solving a certain problem are put together as one set for each property and category, and any two pieces of knowledge among the plurality of knowledge modules are A storage device that stores a knowledge database that has a structure that quantitatively expresses the relationship between modules and the relationship between one knowledge module and a single piece of knowledge belonging to another knowledge module using a numerical value called the degree of connectivity. and a program having a dynamic allocation knowledge utilization function that dynamically extracts or rejects only the knowledge modules necessary in the process of solving a problem based on the degree of connectivity, and A problem-solving device consisting of an input/output device that performs output, and a central processing unit that centrally manages and controls each of the devices and programs mentioned above.