Satellite Constellation Reconfiguration Using Surrogate-Based Optimization

Abstract

The reconfiguration of a constellation with several faulty satellites concerns performance improvements in multiple fields, which can be regarded as a multiobjective optimization (MOO) problem. In the optimization design, it is inevitable to evaluate the constellation performance and providing the reconfiguration strategy thousands of times is time-consuming. To decrease the high computational expense, this paper proposes an accurate and efficient MOO method based on the kriging surrogate model, termed as the surrogate-based MOO (SBMOO) algorithm. A new hybrid refinement method is presented to select infilling samples for updating the surrogate model. Different from the orbital phasing maneuver, the low-thrust reconfiguration strategy is implemented to optimize the transfer trajectory with low fuel consumption, by changing the orbital inclination and right ascension of the ascending node. With the tradeoff between the constellation performance and the uniformity of fuel consumption, the MOO problem of constellation reconfiguration can be investigated and settled by the proposed SBMOO algorithm. The simulations confirm that the preferable constellation reconfigurations are achieved with a low computational expense for optimization and a low fuel cost for orbital transfer.