| Summary: | 碩士 === 國立臺灣大學 === 機械工程學研究所 === 88 === This thesis applies both the symbolic and numeric constraint satisfaction methods to an expert system for mechanical design. First, an object-oriented data model for representing mechanical design problems is established. Within the data model, different kinds of classes and instances are defined to represent the inheritance and composite relationships between components and subsystems of a mechanical design problem. In addition, both symbolic and numerical constraints are also defined to simulate different design limitations for the design problem. Then, an expert system shell based on the constraint satisfaction method is developed. Within the shell, symbolic constraints are converted to rules for the expert system, numerical constraints are transformed to a general numerical optimization problem, and an inference process is carried out to find feasible solutions. A post processor for automatically constructing the solid model of design solutions is also included in the expert system. Finally, detail design processes of several mechanical design examples are given. From these design results, it shows that the proposed method can be used easily and effectively for general mechanical design problems.
|
|---|
