Abstract:

According to the characteristics of complicated simulation computation, large scale design space and nonlinear constraints exsting in the top design of satellite systems, a derivative-free hybrid optimization algorithm is presented, which combines treed Gaussian process (TGP), generalized pattern search (GPS) with the filter algorithm to create a new hybrid optimization algorithm. Design space is partitioned into non-overlapping sub-regions by using the TGP model, and an independent stationary Gaussian process (GP) is bulit in each sub-region to substitute for the practical model. The new “promising” points are generated utilizing the TGP, which are the combination of model predicted values and predict errors. Then, these points are used to guide GPS search in the design space efficiently. The filter algorithm combining GPS is used to deal with nonlinear constraints. The proposed method is applied to the top design of multi-satellites cooperated observation. The results demonstrate that the hybrid algorithm has a good global search ability, which can not only increase the chance of obtaining an optimal solution but also cut down the cost of function evaluations.