优化多星协同观测的改进广义模式搜索算法

多星协同观测可以最大化卫星的整体效能,如何对多星进行部署是一个设计空间大、设计变量多的优化问题.对此,提出了基于Kriging模型的改进广义模式搜索算法.在算法的搜索步,通过代理模型最优和最大期望提高在当前网格内进行选点,避免选择的盲目性；在筛选步,利用代理模型预测筛选集中各点提高的比例并排序,减少不必要的仿真分析.最后,采用该算法对多星部署方案进行优化,通过对比发现,算法性能优于STK-Analyzer,证明了算法的可行性和有效性.

Using improvement generalized pattern search to optimize design multi-satellites' cooperated observation

Multi-satellites’ cooperated observation is an effective way to maximize satellite system performance.How to deploy these satellites is an optimization problem,which involves a huge design space and many design parameters.For this problem,an improved generalized pattern search algorithm,which is based on Kriging model,was put forward.For the search step,two point selection mechanisms were constructed.One is to choose the point with maximal predict value about the surrogate and the other is to choose point with maximal expect improvement.For the poll step,the surrogate was used to predict the expected improvement of the points in the poll set and rank them accordingly.It can reduce the unnecessary simulation analysis.Finally,the proposed algorithm was applied to deploy multi satellites for a given scenario.Compared with the results obtained by STK-Analyzer,the proposed algorithm operated better in performance.The results prove the feasibility and effectiveness of the algorithm.