1992 Fiscal Year Final Research Report Summary
Development of a fuzzy expert system for predicting the probable failure on mountain slopes
| Project/Area Number |
03555111
|
|---|
| Research Category |
Grant-in-Aid for Developmental Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
基礎・土質工学
|
|---|
| Research Institution | Kobe University |
|---|
Principal Investigator |
OKIMURA Takashi Kobe University, Faculty of Engineering, Associate Professor, 工学部, 助教授 (50031125)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
NODA Tsutomu Soil Consultants Co. Ltd. Kobe Office, Chief, 神戸事務所, 所長
KASHIWAYA Kenji Kobe University, Graduate School of Science and Technology, Assistant, 自然科学研究科, 助手 (30161029)
|
|---|
| Project Period (FY) |
1991 – 1992
|
| Keywords | expert system / Fuzzy theory / Dangerous of probable failure / Consulting probable failure / Slope failure / Rock fall |
|---|
| Research Abstract |
An expert system for consulting probable failure of a mountain slope is proposed. PROLOG, an artificial intelligence language was used. The basic probability proposed in the Dempster & Shafer theory was used in order to account for the inaccuracy that accompanies domain-specific knowledge. The basic probabilities were determined with 1)to what degree an item affect the occurrence of surface failure and 2)with what frequency a failure appear for each category of every item. Six fuzzy sets are defined as a goal. To examine weather the proposed system is appropriate, we applied it to an area. Our results show that the dangerous goals are obtainable for failed slopes and the safety goals for non- failed ones. Our results also show that topographical factors greatly influence of the occurrence of surface failure on a mountain slope.
This system is a prototype of an expert system for consulting probable failure. The prototype is examined for a mountain slope by some another experts. As the re
… More
sults, the prototype is modified. Factors of geology, man impacts and countermeasures are supplemented to the model. The basic probability of " probability of "unknown" is changed for many factors in the prototype system. The examination is made with the modified expert system on different mountain slopes. Some problems to be clear in the future are pointed out. These are a knowledge database and a field investigation method.
Furthermore, an expert tool which aid to construct an expert system is employed for the purpose to know what factor affect an appearance of a mountain slope failure. The data are obtained by field investigation of cut, filled and natural slopes along a road in Hyogo prefecture. These data are collected by Hyogo municipal corporation. The data were analyzed by using this tool from regional and geological consideration. The results show that a factor of "history of failure" is very important in some districts, however, in another area, "unstable stones on a slope surface" or "creeping of a surface soil layer". Less
|
