Abstract:
Using generalized logarithmic spirals to approximate low-thrust trajectories, a new
strategy for the design of low-thrust gravity-assist transfers has been developed.
Each transfer leg is defined by a semi-analytic model, and its solution is equivalent
to a hybrid Lambert’s problem. The method is suitable for approximating
both flyby and rendezvous transfer legs. A branch and prune algorithm is used
to generate a collection of initial guesses for further optimization. The analytic
nature of the low-thrust model simplifies the pruning step, since dynamical and
operational constraints (like maximum thrust or total v) can be imposed easily.
The solutions obtained with the global search algorithm can be post-processed,
filtered, and ranked according to various criteria. This is where the versatility of
the method resides, because changing the selection criteria does not require a new
search. Selected candidates are then optimized further, in order to generate actual
low-thrust orbits. Two mission design examples are presented: an asteroid
deflection mission using a kinetic impactor, and a rendezvous mission to Jupiter.
These examples are used to analyze the convergence of the optimization stage, in
particular how far from the optimal solution the initial guesses are.