A web broker for component retrieval in mechanical engineering

✍ Scribed by C.T. Charlton; K.M. Wallace

• Publisher: Elsevier Science
• Year: 2000
• Tongue: English
• Weight: 955 KB
• Volume: 21
• Category: Article
• ISSN: 0142-694X
• DOI: 10.1016/s0142-694x(99)00040-x

✦ Synopsis

W

hen designers set out to create a new product, they seldom start with a blank sheet of paper and proceed top-down from first principles. For much of engineering design, archetypes exist in the form of existing products and standard components. The aim is to adapt these designs to meet new requirements, to increase reliability, and to reduce cost.

The complexity of designing a product as a whole can be reduced by taking fragments from existing designs and combining them, usually with some modifications. Each fragment concerns a particular part of the designed artefact or aspect of the design process. That is, as well as concrete parts and subsystems, it is also possible to transfer knowledge about those designed artefacts and the processes of designing them. In Machine Learning, a similar distinction is made between transformational analogy, in which a concrete solution is reused directly, and derivational analogy, in which the procedure by which a previous solution was obtained is replayed