论文部分内容阅读
The newly developed hybrid and adaptive metamodeling method(HAM) is applied in lightweight design of one car rear frame.To reduce more weight of the system under the premise of the system stiffness, local search scheme is proposed, which is searched in a small subspace constructed by HAM sample point offsetting a small distance beside design space using the metamodels.HAM integrated three different metamodels which are Kriging, radial basis function(RBF) and uadratic function(QF) metamodel in optimization process.In addition, the appropriate metamodel can be selected automatically from the alternative metamodels in solving unknown problems.The employed three metamodels can be adaptively updated in the optimization process and the accuracy in the focused area will be gradually increased.When the optimization by HAM method is done, the local search scheme will be applied in the constructed subspaces for more mass reduction.Application of local search scheme makes the rear frame system reduce more by another 2.18 kg than that just by HAM method.