Multiobjective Intelligent Computer-Aided Design

摘要

The authors present the basis of a knowledge system methodology for multiobjective intelligent computer-aided design (MICAD). A hybrid of formal reasoning and expert systems techniques underlies partitioning the knowledge base between a progressively-acquired preferential component and an a priori acquired operational component. Markedly different representation schemes capture the two components. Dynamic acquisition of the preferential component, utilized in large part to direct the search for improved designs, enables the system to mimic and conform to different users, decision-making styles, and design goals. An example search procedure for MICAD, employing the well-known search algorithm A*, will also discussed. This effort was born with the observation that a particular domain of computer-aided design (CAD) namely packet-switched data communications network design, is ripe for a shift in methodological basis from mathematical programming techniques to knowledge-based and artificial intelligence techniques. (Reprint)