Abstract
This paper proposes an accuracy enhancement method of mini-max type robust optimal design by function suitabilization. While robust optimal design utilizes any model that mediates the mathematical definition of robustness and numerically-computable object models, its fidelity affect the accuracy of solutions. In the case of mini-max type robust optimal design, the characteristic of quadratic response surface for obtaining boundary points is such a concern. This paper proposes a suitabilization filter, which is composed of linear transformation and Fourier series, for the quadratic response surface. The arranged robust optimal design method with the proposed filter is applied to two numerical examples to ascertain its validity.
