Sequential approximate optimization using support vector regression

Support Vector Regression (SVR) is being increasingly used due to its higher accuracy and lower standard deviation than those of existing meta-models. However, while it has been applied to many studies such as the time series prediction and statistical learning theory, SVR has been rarely used for design optimization. In this study, a Sequential Approximate Optimization (SAO) method based on SVR is developed. We adopt the Inherited Latin Hypercube Design (ILHD) for the Design of Experiment (DOE) and the trust region algorithm for the model management technique. Finally, in order to show the accuracy and efficiency of the proposed method, three mathematical problems and a practical design problem are solved and compared a proposed method with SAO using other meta-models such as Kriging, Radial Basis Function (RBF) and Polynomial Regression (PR). The results show that proposed method not only found exact solution at all problems but also performed more efficiently than other SAO methods about 26%similar to 63%.