JPS62256128A - Inference system - Google Patents
Inference systemInfo
- Publication number
- JPS62256128A JPS62256128A JP61099952A JP9995286A JPS62256128A JP S62256128 A JPS62256128 A JP S62256128A JP 61099952 A JP61099952 A JP 61099952A JP 9995286 A JP9995286 A JP 9995286A JP S62256128 A JPS62256128 A JP S62256128A
- Authority
- JP
- Japan
- Prior art keywords
- inference
- rule
- backward
- presupposition
- condition
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 claims abstract description 38
- 238000003745 diagnosis Methods 0.000 abstract description 9
- 238000004321 preservation Methods 0.000 abstract 4
- 150000001875 compounds Chemical class 0.000 abstract 1
- 238000011084 recovery Methods 0.000 description 11
- 238000010586 diagram Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
Abstract
Description
【発明の詳細な説明】
[発明の目的1
(産業上の利用分野)
本発明は、専門家の知識を蓄積し、これを使用して診断
、設計等を支援するエキスパートシステム等の推論方式
に関する。[Detailed Description of the Invention] [Objective of the Invention 1 (Field of Industrial Application) The present invention relates to an inference method such as an expert system that accumulates the knowledge of experts and uses this to support diagnosis, design, etc. .
(従来の技術)
従来推論方式としては、前向き推論方式、および後向き
推論方式がおる。(Prior Art) Conventional inference methods include forward inference methods and backward inference methods.
この前向き推論方式とは、「Xがaならば、yはbであ
る」という規則および「Xがaでおる」という事実を元
に「yはbである」という結論を推論する方式でおる。This forward inference method is a method that infers the conclusion that ``y is b'' based on the rule ``If X is a, then y is b'' and the fact that ``X is a.'' .
また後向き推論方式とは既知でない「yはbである」を
証明したい仮説と児なし、「Xがaならば、yはbであ
る」という規則から「Xはaでおる」ことか成立するか
どうかを調べることによつ゛ て、その仮説の成否を
推論していく方式である。Also, the backward inference method is unrelated to the unknown hypothesis that wants to prove "y is b", and from the rule "If X is a, then y is b", "X is a" is established. This is a method of inferring the success or failure of a hypothesis by examining whether it is true or not.
これらの推論方式をエキスパートシステムの推論方式と
して適用するには、前向き推論方式は計画、設計型エキ
スパートシステムに適していると言われ、また後向き推
論方式は診断型エキスパートシステムに適していると言
われる。When applying these inference methods as inference methods for expert systems, forward inference methods are said to be suitable for planning and design type expert systems, and backward inference methods are said to be suitable for diagnostic type expert systems. .
しかし最近推論によって解こうとする問題の対象分野は
広がり、「この問題は何々型だから、この推論方式を採
用する」というように簡単に決めることができなくなっ
ている。むしろこれら2つ゛ の推論方式を組合せる
ことによって、両方の特徴を生かした推論を行わなけれ
ばならない場合が多い。However, recently the scope of problems to be solved by inference has expanded, and it is no longer possible to simply decide, ``This problem is of such-and-such type, so I will adopt this inference method.'' Rather, it is often necessary to combine these two inference methods to perform inference that takes advantage of the characteristics of both.
例えばおる機械の故障診断を行った後、その診断結果を
前提条件として修理のための対策方法を推論したり、ざ
らにその対策方法を前提条件として必要な部品、工具の
種類や数量を推論したい場合がある。この場合故障診断
を後向き推論で行い、その仙を前向き推論によって実施
することが望ましい。For example, after diagnosing the failure of a machine, you can use the diagnosis results as a prerequisite to infer countermeasures for repair, or roughly use the countermeasures as a prerequisite to infer the types and quantities of necessary parts and tools. There are cases. In this case, it is desirable to perform fault diagnosis using backward reasoning, and then perform the fault diagnosis using forward reasoning.
このようなときに独立した前向き推論方式と後向き推論
方式とを組合せて上記推論を実現しようとすると、ルー
ル表現を統一することが困難であり、多大な開発工数が
費され、全体としての推論の処理効率が悪くなるという
問題がめった。In such a case, if you try to realize the above inference by combining the independent forward inference method and backward inference method, it will be difficult to unify the rule expressions, a large amount of development man-hours will be spent, and the overall inference will be The problem of poor processing efficiency was common.
(問題点を解決するための手段)
従来の推論方式では、推論方式が固定され、用途に応じ
て柔軟に対応することができないという問題点がめった
。(Means for solving the problem) Conventional inference methods often have the problem that the inference method is fixed and cannot be flexibly adapted to suit the application.
本発明はこの事情に対処してなされたもので、前向き推
論および後向き推論を柔軟に実行することのできる推論
方式を提供することを目的とする。The present invention was made in response to this situation, and it is an object of the present invention to provide an inference method that can flexibly execute forward inference and backward inference.
[発明の構成]
(問題点を解決するための手段)
本発明は上記問題点を解決するために、前提条件と推論
規則とから推論を行う推論方式において、仮説の成立を
推論規則から検証する後向き推論手段と、前提条件およ
び推論規則から推論を行う前向き推論手段とを設け、各
段階ごとの中間推論結果を得ながら多段推論を行い、所
定の段階までの中間推論結果は後向き推論手段による後
向き推論により得、それ以後の段階の中間IW論結果は
前向き推論手段による前向き推論により得るようにした
ものである。[Structure of the Invention] (Means for Solving the Problems) In order to solve the above problems, the present invention verifies the establishment of a hypothesis from the inference rules in an inference method in which inferences are made from preconditions and inference rules. A backward inference means and a forward inference means that performs inference from preconditions and inference rules are provided, multi-stage inference is performed while obtaining intermediate inference results at each stage, and the intermediate inference results up to a predetermined stage are retrospectively inferred by the backward inference means. Intermediate IW theory results at subsequent stages are obtained by forward reasoning using forward reasoning means.
(作用) 従って推論を以下のようにして行うことができる。(effect) Therefore, inference can be made as follows.
すなわち、まず後向き推論手段による後向き推論によっ
て仮説を検証しながら所定の中間推論結果を得る。That is, first, a predetermined intermediate inference result is obtained while verifying a hypothesis by backward inference by the backward inference means.
次にこの中間推論結果を前提条件として前向き推論手段
により前向き推論を実行する。Next, forward inference is executed by forward inference means using this intermediate inference result as a precondition.
さらにこの中間推論結果を別の目標を)qるための前提
条件と見なして前向き推論を行ない、多段推論を実行す
る。Further, this intermediate inference result is regarded as a precondition for determining another goal, and forward inference is performed to perform multi-stage inference.
(実施例) 以下、図面を参照しながら本発明の詳細な説明する。(Example) Hereinafter, the present invention will be described in detail with reference to the drawings.
第1図は本発明の一実施例に係わる推論装置の4g成を
示すブロック図である。FIG. 1 is a block diagram showing a 4G configuration of an inference device according to an embodiment of the present invention.
この推論装置は、前向き推論が実行される前向き推論部
1と、前向き推論部1で使われる条件の成否が検証され
る後向き推論部2と、I F−THEN形式のルールが
保存されるルール保存部3とより構成されている。This inference device includes a forward inference unit 1 in which forward inference is executed, a backward inference unit 2 in which the success or failure of conditions used in the forward inference unit 1 is verified, and a rule storage unit in which rules in IF-THEN format are stored. It consists of Section 3.
さらに前向き推論部1は、前提条件もしくは直前の推論
結果が保存される前提条件保存部11と、前提条件保存
部11に保存されている前提条件を条件部における条件
の1つとしているルールをルール保存部3から捜し出す
ルール実行部12と、サーチされたルールを実行するル
ール実行部13と、実行の結果得られた結論か推論の目
標であるかどうかを判定する結論判定部14とから構成
されている。Further, the forward inference section 1 stores a precondition storage section 11 in which the precondition or the previous inference result is stored, and a rule that sets the precondition stored in the precondition storage section 11 as one of the conditions in the condition section. It is composed of a rule execution unit 12 that searches the storage unit 3, a rule execution unit 13 that executes the searched rule, and a conclusion determination unit 14 that determines whether the conclusion obtained as a result of the execution is the target of the inference. ing.
次に第2図を用いてこの推論装置の動作を説明する。Next, the operation of this inference device will be explained using FIG.
第2図はこの推論装置によって実行される改障診断lW
論過程を示す図である。Figure 2 shows the failure diagnosis lW executed by this reasoning device.
FIG.
この故障診断ては、まず故障原因か同定され、次にその
故障原因に対する復旧方法が求められ、さらにこの復旧
方法に必要な資材が求められる。In this fault diagnosis, first the cause of the failure is identified, then a recovery method for the cause of the failure is determined, and materials necessary for this recovery method are also determined.
まず故障原因を仮定することによって後向き推論Aが起
動され、この故障原因を検証するためにその条件となる
事象の成否の評価が繰返される。First, by assuming the cause of the failure, backward reasoning A is activated, and in order to verify the cause of the failure, evaluation of the success or failure of the event serving as the condition is repeated.
以上の操作によって故障原因が後向き推論Aにより同定
されると、次にこの故障原因を前提条件として復旧方法
の推論が起動される。この復旧方法の推論は前向き推論
Bと、この前向き推論Bにおいて条件部の条件中で成否
がわからないものを検証する後向き推論Cとによって推
論が進められていく。When the cause of the failure is identified by backward reasoning A through the above operations, the inference of a recovery method is then activated using this cause of failure as a precondition. The inference of this recovery method is advanced by forward inference B and backward inference C, which verifies conditions in the condition part of this forward inference B that are not known whether they are successful or not.
ずなわら同定された故障原因は、前提条件保存部11に
保存される。するとルール保存部3に保存されているル
ールのうち、その条件部の一部が保存された故障原因と
一致するものがルールサーチ部12によって選択される
。The identified cause of failure is stored in the prerequisite storage unit 11. Then, among the rules stored in the rule storage section 3, the rule search section 12 selects one whose condition part partially matches the stored failure cause.
続いてルール実行部13において選択されたルールが実
行されるわけであるが、このルールの条件部に故ド9原
因以外の事象が存在する場合は、この事象の成否か後向
き推論部2によって検証される。Next, the selected rule is executed in the rule execution unit 13, but if an event other than the cause of the failure exists in the condition part of this rule, the backward reasoning unit 2 verifies whether this event is successful or not. be done.
そして条件部のすべての条件か成立すると、このルール
の結論がこの前向き推論の目標である復旧方法であるか
どうかが結論判定部14で判定され、目標の復旧方法で
あればこの復旧方法を前向き推論Bの結論とする。When all the conditions in the conditional part are satisfied, the conclusion determining part 14 determines whether the conclusion of this rule is the recovery method that is the goal of this forward inference, and if it is the target recovery method, this recovery method is This is the conclusion of Inference B.
もし目標の復旧方法でなければ、この前向ぎ推論の結果
を前提条件保存部11に保存し、上記の操作を繰返す。If it is not the target recovery method, the result of this forward inference is stored in the precondition storage unit 11, and the above operation is repeated.
以上のようにして故障原因より復旧方法を推論すること
ができる。As described above, a recovery method can be inferred from the cause of the failure.
同様にしてこの復旧方法を前提条件として必要な資材の
推論、すなわち前向き推論りおよび後向き推論Eも引き
続き実行することができる。Similarly, with this recovery method as a prerequisite, the necessary material inferences, ie, forward inference and backward inference E, can be continued.
[発明の効果]
以上述べたように本発明によれば、後向き推論および前
向き推論を組み合せた多段推論を行うことができるので
、復旧方法も含めた故障診断等の複合的な問題を解決す
ることができ、多目的かつ高性能なエキスパートシステ
ムを横築することができる。[Effects of the Invention] As described above, according to the present invention, multi-stage inference that combines backward inference and forward inference can be performed, so complex problems such as failure diagnosis including recovery methods can be solved. It is possible to horizontally build a multi-purpose and high-performance expert system.
第1図は本発明の一実施例に係わる推論装置の構成を示
すブロック図、第2図は同実施例によって実行される故
障診断推論過程を示す図である。
1・・・・・・・・・前向き推論部
2・・・・・・・・・後向き推論品
出願人 株式会社 東芝
代理人 弁理士 須 山 佐 −
第1図
第2図FIG. 1 is a block diagram showing the configuration of an inference device according to an embodiment of the present invention, and FIG. 2 is a diagram showing a fault diagnosis inference process executed by the embodiment. 1...Forward reasoning section 2...Backward reasoning Product applicant Toshiba Corporation Representative Patent attorney Sasu Suyama - Figure 1 Figure 2
Claims (4)
おいて、仮説の成立を前記推論規則から検証する後向き
推論手段と、前記前提条件および前記推論規則から推論
を行う前向き推論手段とを設け、所定の段階までの中間
推論結果は前記後向き推論手段による後向き推論から得
、それ以後の段階の中間推論結果は前記前向き推論手段
による前向き推論から得る多段推論を行うことを特徴と
する推論方式。(1) In an inference method that makes inferences from preconditions and inference rules, there is provided a backward inference means for verifying the establishment of a hypothesis from the inference rules, and a forward inference means for making inferences from the preconditions and the inference rules, An inference method characterized in that intermediate inference results up to a predetermined stage are obtained from backward inference by the backward inference means, and intermediate inference results at subsequent stages are obtained from forward inference by the forward inference means.
られた中間推論結果をさらに前提条件として前向き推論
を継続する特許請求の範囲第1項記載の推論方式。(2) The inference method according to claim 1, wherein the forward inference means continues the forward inference using the intermediate inference result obtained by the immediately preceding forward inference as a precondition.
ける条件部の条件が前提条件に含まれていない場合この
条件を仮説として後向き推論手段を起動する特許請求の
範囲第1項もしくは第2項記載の推論方式。(3) The forward inference means is configured to activate the backward inference means using this condition as a hypothesis when the condition in the condition part of the inference rule is not included in the prerequisites at the time of forward inference. reasoning method.
のみ起動される特許請求の範囲第1項ないし第3項のい
ずれか1項記載の推論方式。(4) The inference method according to any one of claims 1 to 3, wherein the forward inference means is activated only by the immediately preceding intermediate inference result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP61099952A JPS62256128A (en) | 1986-04-30 | 1986-04-30 | Inference system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP61099952A JPS62256128A (en) | 1986-04-30 | 1986-04-30 | Inference system |
Publications (1)
Publication Number | Publication Date |
---|---|
JPS62256128A true JPS62256128A (en) | 1987-11-07 |
Family
ID=14261034
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
JP61099952A Pending JPS62256128A (en) | 1986-04-30 | 1986-04-30 | Inference system |
Country Status (1)
Country | Link |
---|---|
JP (1) | JPS62256128A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0296236A (en) * | 1988-09-30 | 1990-04-09 | Omron Tateisi Electron Co | Fuzzy computer |
JPH02156331A (en) * | 1988-12-08 | 1990-06-15 | Hitachi Ltd | Method and device for aiding judgement |
JPH0393427A (en) * | 1989-09-02 | 1991-04-18 | Tokyo Electric Power Co Inc:The | Determination of operating sequence of switchgear in power station |
-
1986
- 1986-04-30 JP JP61099952A patent/JPS62256128A/en active Pending
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH0296236A (en) * | 1988-09-30 | 1990-04-09 | Omron Tateisi Electron Co | Fuzzy computer |
JPH02156331A (en) * | 1988-12-08 | 1990-06-15 | Hitachi Ltd | Method and device for aiding judgement |
JPH0393427A (en) * | 1989-09-02 | 1991-04-18 | Tokyo Electric Power Co Inc:The | Determination of operating sequence of switchgear in power station |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7433854B2 (en) | Backward chaining with extended knowledge base network | |
De Giacomo et al. | Imperfect-Information Games and Generalized Planning. | |
JPS62256128A (en) | Inference system | |
US4899290A (en) | System for specifying and executing protocols for using iterative analogy and comparative induction in a model-based computation system | |
JPH047743A (en) | Inference system | |
Lorang et al. | Speeding-up continual learning through information gains in novel experiences | |
Liu et al. | Active policy improvement from multiple black-box oracles | |
JPS624663A (en) | Fault diagnosing apparatus for vehicle | |
Girgin et al. | Learning by automatic option discovery from conditionally terminating sequences | |
JP3344540B2 (en) | Expert system | |
JPS63239528A (en) | Production system | |
JPS62198930A (en) | Reasoning method | |
Kapur et al. | Proving equivalence of different axiomatizations of free groups | |
Girgin et al. | State Similarity Based Approach for Improving Performance in RL. | |
JPH07334371A (en) | Inference device | |
JPH01224842A (en) | Inferential control system for knowledge-based system | |
JPH0650467B2 (en) | Inference processor | |
JPH0232426A (en) | Code preparing and processing method for compiler | |
JPH03218536A (en) | Diagnostic system for data processor | |
JPH04195631A (en) | Inference device | |
JPH0991036A (en) | Plant monitoring device | |
JPH01199236A (en) | Inference system | |
Paramasivam et al. | Hybrid Chaining Inference Technique | |
JPS63220327A (en) | Knowledge technology tool | |
JPH01124030A (en) | Inference method in knowledge utilizing information processing system |