Geometric reasoning for knowledge-based parametric design using graph representation

Abstract

A promising approach to parametric design is based on the knowledge-based technique. However, one of the major drawbacks in this approach is that the inference process is computationally expensive to be applied to the interactive and intelligent CAD systems. This paper presents a new approach to geometric reasoning for knowledge-based parametric design using graph representation to improve the inference process. The geometric reasoning procedure consists of three steps: (1) representing a well-constrained design model and geometric rules into graphs; (2) selecting appropriate subgraphs from the design graph which may be used to induce new facts; and (3) selectively searching for the rule graphs having the same keys as the model subgraphs. The proposed approach is simple in concept, yet realizes significant inference time reduction. The concept presented here has been implemented on an IRIS Indigo workstation, and some implementation results are given to show the efficiency of the proposed approach. Copyright (C) 1996 Elsevier Science Ltd