Swarm Optimization of Lunar Transfers from Earth Orbit with Operational Constraints

Abstract

Due to the frequent launches of primary satellites into geosynchronous transfer orbits, lunar transfer from an initial geosynchronous transfer orbit represents an attractive option for secondary payloads with lunar mission objectives. The development of advanced launch vehicle upper stages with low boil off rates and restartable engines further enhances the potential of geosynchronous transfer orbits to serve as starting points for lunar transfers. Lunar trajectory design from arbitrarily oriented Earth orbits involves complex tradeoffs that can be successfully addressed with particle swarm optimization. An important mission design consideration is radiation dose, since coasting in a geosynchronous transfer orbit involves repeated passes through the Van Allen radiation belts. In this investigation, a mixture density neural network is trained on the state-of-the-art radiation environment models to provide a surrogate model capable of making probabilistic estimates of radiation dosage. The network is combined with the particle swarm method to optimize the coasting and transfer arcs of a lunar trajectory. The end-to-end design of a trajectory from geosynchronous transfer orbit into a specified lunar orbit is demonstrated by modeling the optimized trajectory in a high-fidelity orbit propagator, including B-plane targeting to overfly a desired lunar landing site.