Knowledge acquisition and development of an expert system for type synthesis of *mechanisms

Biwu Yang, University of Rhode Island

Abstract

Methodologies are developed and implemented in LISP to develop an integrated expert system, called DOMES (Design Of Mechanisms by an Expert System), for type synthesis of mechanisms. These methodologies involve techniques for acquiring practical design knowledge from human designers such that the knowledge base of the system can be dynamically expanded, which is an important feature toward a useful expert system, and techniques of conflict resolution when applying a set of design criteria so that the time of problem solving is reduced. Several modified modules, such as central Expert Designer and Knowledge Engineer Module, significantly enhance the capability of DOMES and improve its performance to solve realistic design problems.^ DOMES is equipped with a new Knowledge Engineer module based on natural language understanding techniques. A human designer can add his/her expertise, expressed in English sentences, and the Knowledge Engineer module translates this knowledge into Rejection Functions in LISP code which DOMES can understand and store into the knowledge base of the system. Thus, a designer does not have to be an expert in graph theory, AI, expert systems development, or LISP programming to develop functions for criteria which are not presently in the system.^ The results using the Knowledge Engineer module show that it can effectively acquire knowledge from human designers and, in addition, it has the capability to acquire new concepts through a "user defining interaction" procedure. The expert system equipped with such a feature could become a powerful tool for a designer at the conceptual stages of the design process, where creativity is essential. The concepts and implementation of this module can also be extended to other knowledge based systems to serve as a knowledge acquisition facility or, in general, serve as an intelligent man-machine interface.^ The revised version of the central Expert Designer module provides more flexible yet thorough procedures dealing with design problems. The consideration in these procedures is based on experience from practical mechanism design projects. Thus, the system developed is useful in solving realistic design problems. ^

Subject Area

Engineering, Mechanical|Artificial Intelligence

Recommended Citation

Biwu Yang, "Knowledge acquisition and development of an expert system for type synthesis of *mechanisms" (1991). Dissertations and Master's Theses (Campus Access). Paper AAI9203003.
http://digitalcommons.uri.edu/dissertations/AAI9203003

Share

COinS