107 SAとMVFOSMによるロバスト最適化

抄録

The robust optimization methods generally apply the two-stage design method which the Taguchi method and the Design For Six Sigma (DFSS) represented. In the two-stage design, the first stage of DFSS is optimized by nominal design variables. At its second stage, initial design values of the robust optimization are defined with optimal solution of previous stage, and the optimization is performed. If robust optimal solution doesn't exist at neighborhood of its initial design values, the two-stage design method doesn't function effectively. Furthermore, a difficulty of the robust optimization increases, if design problem contains discrete or mixed variables. In this paper, to overcome its difficulty the robust optimization method is proposed by using SA and MVFOSM based on one-stage design method. The proposed method is applied to some test problems to check its performance. Furthermore, the validity of the proposed method is discussed through the optimization of space satellite system.