Investigations of temperature loads transfer for multidisciplinary design of turbine blades

Hu Dianyin,Yang Junjie,Wang Rongqiao

DOI: https://doi.org/10.3969/j.issn.2095-2783.2012.08.004

2012-01-01

Abstract:Due to differences in computing models and computing grids used,the transfer of results between disciplines must rely on interpolation algorithms.In this paper,approximate models are applied to solve the loads transfer for multidisciplinary optimization(MDO) of turbine blades.By comparing the efficiency and accuracy of various approximate models during interpolation in the temperature loads transfer,including the polynomial model,Kriging model,radial basis function(RBF) model and support vector machine(SVM) model,it is found that the SVM model is suitable only for small samples,thus it cannot be used for the MDO design of turbine blades;the polynomial model suits for large sample cases;when the sample size is moderate,the interpolation results based on the Kriging model and the RBF model are better.