Multicriteria robust design of electronic devices using enhanced genetic algorithms

The demand for high performance electronic devices is steadily pushing industry toward off-line quality control, that is, toward products and manufacturing processes whose performance is robust or insensitive to environmental or other uncontrollable factors. When the design is simple, Taguchi methods may do the job here quite effectively. In this paper we tackle some exceptionally difficult multi-objective robust design problems and also those involving discrete design factors. Such problems are beyond either the classical Taguchi method or the response surface-based methods put forth recently. We propose a methodology based on enhanced genetic algorithms that directly produces Pareto-optimal robust designs. We re-visit a bi-criteria electronic filter design problem to illustrate this new procedure. No other technique appears to exist in the literature that produces such designs.