Search

The problem of earth observation satellite system (EOSS) optimization is characterized by computationally expensive, large scale design space, nonlinear constraints. According to different phases of simulation based optimization, a framework of experimental design is built. In the initial phase, orthogonal design is borrowed to rank related system variables. Then, uniform design or Latin Hyper-cube design is adopted to create simulation plan. Finally, robust design is used to ensure the solution be of good robustness. By using the proposed framework, the scale of simulation can be greatly cut down, simulation time be badly shortened and simulation efficiency be largely improved. The most important should be that the quality of solution can be guaranteed ,and finally can be proved by the example constructed in the final part. [ABSTRACT FROM AUTHOR]