Adaptive multi-segment pseudospectral sequential convex programming for satellite cluster reconfiguration trajectory optimization

This paper presents an adaptive pseudospectral sequential convex programming method aimed at enhancing computational efficiency and safety in large-scale satellite cluster reconfiguration trajectory optimization. First, a multi-segment pseudospectral discretization technique is combined with successive linearization to transform the nonconvex optimal control problem into a series of convex subproblems. Second, an adaptive segmentation strategy based on collision warnings divides the entire orbital transfer process into collision-warning and non-collision-warning phases, with varying grid densities applied. This strategy improves satellite safety without increasing the total number of discretization points. Third, a filtering strategy is introduced to accurately identify the collision avoidance objects and periods, eliminating a large number of inactive collision avoidance constraints. This reduces the complexity of the convex subproblems and alleviates the computational burden. Finally, numerical simulation results demonstrate that the proposed method outperforms the traditional sequential convex programming and GPOPS methods in terms of efficiency, accuracy, and safety in scenarios involving 2 to 100 satellites. For the 8-satellite case, the proposed method improves computational efficiency by 99% over GPOPS and 65% over trapezoidal discretization sequential convex programming.