Fault Analysis and Diagnostic System by Case-based Reasoni(Application of Analogy in Artificial Intelligence)
| Project/Area Number |
04805024
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
機械力学・制御工学
|
|---|
| Research Institution | KYOTO UNIVERSITY |
|---|
Principal Investigator |
ADACHI Norihiko (1993) Kyoto Univ.Fac.of Eng.Professor, 工学部, 教授 (30026108)
石田 好輝 (1992) 京都大学, 工学部, 助手 (80159748)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
DOUZONO Hiroshi Kumamoto Univ.Fac.of Eng.Lecturer, 工学部, 講師 (00217613)
ISHIDA Yoshiteru Kyoto Univ.Fac.of Eng.Assistant Professor, 工学部, 助手 (80159748)
|
|---|
| Project Period (FY) |
1992 – 1993
|
| Project Status |
Completed (Fiscal Year 1993)
|
| Budget Amount *help |
¥1,700,000 (Direct Cost: ¥1,700,000)
Fiscal Year 1993: ¥300,000 (Direct Cost: ¥300,000)
Fiscal Year 1992: ¥1,400,000 (Direct Cost: ¥1,400,000)
|
|---|
| Keywords | Artigicial Intelligence / Case-based Reasoning / Large-scale systems / Fault Diagnosis / Inference Systems / Knowledge Acquisition / Process Diagnosis / プロセス診断 / 事故解析 / 診断システム |
|---|
| Research Abstract |
In case-based reasoning, the past cases are stored as empirical knowledge to solve the current problems. First, the most similar cases are searched, and then solve the current problems using the anology of the similar cases. We developed an intelligent case retrieval and reasoning system for the fault analysis and diagnosis of processing plants. Since similar accidents happened repeatedly by the similar causes, this retrieval and reasoning system may be important to prevent similar accidents using the knowledge of the past accidents. In this research project, we obtained the following results. (1)First, we developed a memory for case knowledge-base which allows storing cases in orderly fashion. We also developed a fast algorithm to retrieve similar cases in the memory. (2) We developed a model to classify the cases by the pattern of flow structure in the plant. The model also allows to represent the operations by human operator. Thus, the model can include faults by human operation error. (3) The reasoning system above mentioned is tested against a fault simulator. By this test, one can know how much qualitatively diferent cases are needed to for the system to reasonably react the human questions. It also suggests the strategy for collecting the cases for better performances. (4)We also developed a transforming system whitch transforms the similar case so that the case can be used for the current problem. The system include several heuristics for the pattern of transformations.
The aim of this research is to develop an intelligent and flexible retrieval system that allows more than usual search by keywords. We developed the system that can carry out the analogical reasoning rather than search, and we evaluated that such system is useful for the prevention of similar accidents in processing plants.
|
Report
(3 results)
Research Products
(23 results)
