Title: A neural network-based computer aided design tool for automotive form design

Authors: Yu-Ming Chang, Hung-Yuan Chen

Addresses: Department of Industrial Design, National Cheng Kung University (NCKU), No.1, Dasyue Road, East District, Tainan City 701, Taiwan, ROC. ' Department of Industrial Design, National Cheng Kung University (NCKU), No.1, Dasyue Road, East District, Tainan City 701, Taiwan, ROC

Abstract: Although the functionality and performance are both important aspects in vehicle design, an automotive form is a crucial factor in determining a consumer|s image perception and purchase decision. Hence, an effective tool for designing successful automotive form was suggested. Based on Kansei Engineering principles, the relationship between the profile characteristics and the consumer|s image perception is established using a Back-Propagation Neural (BPN) network. A Computer Aided Design (CAD) tool, which uses the trained BPN to predict the consumer perception of an automotive profile expressed in the form of a numerical definition, is constructed using Visual Basic software. The performance of the CAD prototype tool is verified by comparing its predictions to the actual consumer perception evaluations. A good similarity is identified between the two sets of results. Therefore, the developed tool provides designers with powerful means of creating automotive designs from a consumer|s image perception perspective.

Keywords: back-propagation neural networks; computer aided design; CAD; Kansei engineering; numerical definition; automotive form design; vehicle design; automotive profile.

DOI: 10.1504/IJVD.2007.012300

International Journal of Vehicle Design, 2007 Vol.43 No.1/2/3/4, pp.136 - 150

Published online: 04 Feb 2007 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article