LNCS Homepage
CD ContentsAuthor IndexSearch

Optimized Design of MEMS by Evolutionary Multi-objective Optimization with Interactive Evolutionary Computation

Raffi Kamalian1, Hideyuki Takagi2, and Alice M. Agogino1

1University of California, Berkeley, CA, 94720, USA
raffi@me.berkeley.edu
aagogino@me.berkeley.edu

2Kyushu University, Fukuoka, 815-8540, Japan
takagi@design.kyushu-u.ac.jp

Abstract. We combine interactive evolutionary computation (IEC) with existing evolutionary synthesis software for the design of micromachined resonators and evaluate its effectiveness using human evaluation of the final designs and a test for statistical significance of the improvements. The addition of IEC produces superior designs with fewer potential design or manufacturing problems than those produced through the evolutionary synthesis software alone as it takes advantage of the human ability to perceive design flaws that cannot currently be simulated. A user study has been performed to compare the effectiveness of the IEC enhanced software with the non-interactive software. The results show that the IEC-enhanced synthesis software creates a statistically significant greater number of designs rated best by users.

LNCS 3103, p. 1030 ff.

Full article in PDF


lncs@springer.de
© Springer-Verlag Berlin Heidelberg 2004