A domain knowledge based search advisor for design problem solving environments

Abstract

This paper proposes an automatic and adaptive Domain Knowledge Based (DKB) Search Advisor for use with Design Exploration Systems (DES)—a form of design Problem Solving Environment (PSE). The advisor contains domain knowledge of search routine performance on design problems built using a knowledge modelling methodology. These help designers working on complex engineering problems to decrease the cost of design-space search and improve the quality of the resulting designs. This paper introduces this field, beginning with a view of some of the problems and inefficiencies of present design processes. This is followed by the description of a knowledge modelling methodology that may be used to build knowledge models of search routine performance on design domains. One focus of the paper is the use of machine learning to automate the process of knowledge discovery. The practicability of the DKB Search Advisor is then demonstrated with a case study taken from the aircraft wing design domain. The results presented help provide insights into the strengths and weaknesses of various optimization routines. More importantly, they also illustrate that an advisor containing knowledge of search routine performance on design domains can support design engineers in their search activities. The Search Advisor helps to decrease the cost of aircraft wing design search while at the same time increasing the quality of the resulting designs.

More information

Identifiers

ORCID for A.J. Keane: orcid.org/0000-0001-7993-1569

Catalogue record

Export record

Altmetrics