Title: Streamlining automotive product development using neural networks

Authors: Jennifer L. Johrendt, Peter R. Frise, Mohammed A. Malik

Addresses: Mechanical, Automotive, and Materials Engineering, University of Windsor, 401 Sunset Avenue, Windsor, Ontario, N9B 3P4, Canada. ' Mechanical, Automotive, and Materials Engineering, University of Windsor, 401 Sunset Avenue, Windsor, Ontario, N9B 3P4, Canada. ' Chrysler Canada Inc., CIMS 242-01-02, P.O. Box 1621, Windsor, Ontario, N9A 4H6, Canada

Abstract: To remain competitive in the global market, automotive companies scrutinise product development processes for time and cost savings incurred bringing new vehicles to market. As a result, virtual durability simulation can be utilised early in the design process to reduce the number of costly physical prototypes, assuming that high fidelity models are used. The compromise between model accuracy and computational efficiency presents a challenge that will be addressed by the authors. Neural networks, as computationally efficient mathematical models, will be shown as a viable tool for development of high fidelity models of nonlinear hysteretic components within a virtual durability simulation.

Keywords: automotive product development; neural networks; durability simulation; automobile industry; vehicle design; product design; model accuracy; computational efficiency; nonlinear hysteretic components.

DOI: 10.1504/IJVD.2008.020878

International Journal of Vehicle Design, 2008 Vol.47 No.1/2/3/4, pp.19 - 36

Available online: 22 Oct 2008 *

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