A constraint based maximin sampling using correlation matrix for kriging models

Abstract

Approximation models(also known as metamodels) have been widely used in the field of engineering design to make computationally feasible design optimization computationally achievable. The accuracy of approximation models related sampling strategies used. This paper focuses on find the optimal point more quickly than other approximation models. We propose a constraint based maximin sampling using correlation matrix which hybrids maximinsampling with correlation matrix. The reason why this method use approximation model is that applies penalty function to maximinsampling. As an example, we used a bipolar plate in vanadium redox flow battery to reduce the bipolar plate weight.