Abstract
This paper proposes a new design optimization by integrating an evolutionary search and a cumulative function approximation. While an evolutionary search has an ability of global search even under multi-peaks, rugged natures, etc., it requires much computational cost. While a response surface technique can save computational cost for complicated design problems, the fidelity of solution is affected by density of samples. The proposed method simultaneously performs these two methods. In its early phase, it searches roughly but globally approximated surfaces with relatively small number of samples, and in its later phase, it searches intensively around promising regions over response surfaces enhanced with additional samples. It is implemented with a real-coded genetic algorithm and a Voronoi diagram based cumulative approximation, and it is applied to several numerical example problems.
