Research of MYCIN-Based River Disaster Prevention Expert System

• Main Authors: Yen-Ni Chao, 趙燕呢
• Other Authors: Meng-Syun Sie
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/58480073970290183162

碩士 === 國立中興大學 === 土木工程學系所 === 96 === Thanks to broad range of river reconstruction, and various management methods and construction types, engineers generally have to be equipped with more professional knowledge and pragmatic experiences (e.g., choosing appropriate methods at appropriate situations) to "efficaciously" solve the river reconstruction problems. Under such circumstances, lots of fully experienced engineers have been trained over these years; however, their precious engineering knowledge is hardly to be systematically kept due to their leaving and retirement. This makes those novices devoid of long-term training unable to deal with gradually complicated river reconstruction problems; therefore, how to pass down predecessors’ knowledge and experiences thus turns out to be an urgent task. Employing MYCIN created by Stanford University as theoretical foundation, this study develops a series of MYCIN-Based River Disaster Prevention Expert System (MyRes). Through constant interviews with experienced engineers and experts, their knowledge and experiences in this regard have been partly excerpted and categorized and “If-Then” rules toward more than eight hundreds river reconstructions are set up. Specifically, MyRes adopts "Certainty Factor" to deal with the incomplete determinacy between the rules and facts. Meanwhile, this study improves and expands MYCIN rules from “many-to-one” restriction to "many-to-many" rule and adds up "Forward Chaining Inference Engine" to speculate the correlation between each rule and facts generated. Finally, disastrous descriptive factors are divided into twelve categories and stored at expert-knowledge-rule system. To facilitate its application, MyRes has been expanded to Web-edited expert system. Through empirical results, MyRes is proven to efficiently simulate interactive circumstances with experts and to assist engineers by offering appropriate methods to solve river reconstruction problems.