Approximate structural optimization using kriging method and digital modeling technique considering noise in sampling data

This paper describes a combination approach of a digital finite element modeling technique and the Kriging method for structural optimization. Since the digital modeling technique includes some inaccuracies in a modeling process, applicability of the Kriging method to noisy data is investigated. An estimated surface generated by the conventional Kriging method will be wavy and not appropriate to approximate optimization with noisy data. Therefore, a new Kriging-based approach is proposed. The new class of the Kriging method and a digital modeling technique is applied to an eigenfrequency optimization of a structure. Numerical examples illustrate a validity and effectiveness of the proposed method.