![]() |
|||
![]()
|
![]() |
![]() |
![]() |
Chapter 21
|
1. | Introduction | |
2. | Background | |
3. | Techniques I: Problem Definition | |
3.1. | Functions | |
3.2. | Attributes, Parts, and Catalogs | |
3.3. | Constraints | |
3.4. | Properties of a Design | |
3.5. | Specifying a Design Problem | |
4. | Techniques II: Current Methods | |
4.1. | CSP-Based Methods | |
4.2. | Distributed Agent Methods | |
5. | Research Issues | |
6. | Trends and Summary | |
Acknowledgments | ||
References |
In addition to being the central activity of engineering, design is also a central issue in artificial intelligence research. Indeed, recent advances in the field of artificial intelligence (AI), particularly symbolic representation and related problem-solving methods, including knowledge-based (expert) systems, have provided significant opportunities to clarify and articulate concepts of design so as to lay a better framework for design research and design education (Dym and Levitt, 1991; Dym, 1994). Inasmuch as there is within AI a substantial body of material concerned with understanding and modeling cognitive processes, and since the level of articulation in this work transcends in many ways the common vocabulary of engineering design, we here adapt and appropriate some of the vocabulary and paradigms of AI to enhance our understanding of configuration design.
Designs can be characterized as falling into one of three broad classes: creative, the rarest, in which case completely new products are found; variant, wherein we understand the sources of design knowledge quite well, but we are not entirely sure how that design knowledge should be applied; and routine, for which we know exactly how to complete a design. Configuration selection refers to the task of assembling or organizing a known set of parts into a specified architecture, and it would be considered as routine. However, configuration selection problems may present significant challenges because the search space can be quite large if there are many feasible combinations of large numbers of parts. The pioneering R1/XCON system for configuring VAX computer systems for assembly incorporated just such a configuration selection task (McDermott, 1981).
The completion of configuration design tasks requires that a known set of parts be fit within an architecture that has not been specified in advance. As a result, the architecture and some parts have to be designed as part of the organization process. There may be a large number of design parameters, and there may be significant spatial reasoning issues that need to be addressed during the configuration of the architecture. The PRIDE system for the design of paper paths in copier systems is considered the pioneering complex configuration design system (Mittal, Dym, et al., 1986).
Many configuration selection and design systems have been built and described in the last 15 years because the tasks they encapsulate are important in both engineering design and manufacturing. Reviews of such systems can be found in Dym and Levitt, 1991; Dym, 1994; and Darr, 1996.
Previous | Table of Contents | Next |
![]() |
|
Use of this site is subject certain Terms & Conditions. Copyright (c) 1996-1999 EarthWeb, Inc.. All rights reserved. Reproduction in whole or in part in any form or medium without express written permission of EarthWeb is prohibited. Please read our privacy policy for details. |