A Survey of Near-Earth Asteroids for Low-Thrust Round-Trip Missions

The objective of this study is to perform a broad low thrust (LT) round-trip accessibility analysis for near-Earth asteroids (NEAs). The impulsive missions to NEAs have been investigated in several studies from various perspectives, while NEAs' low-thrust missions have not been properly investigated due to the complexity of LT trajectory design. A Deep Neural Network (DNN) classifier is constructed and trained to predict the feasibility of low thrust transfers between Earth and NEAs. This model has a prediction accuracy of 98%, and it is used for filtering out infeasible transfers and enhance the search efficiency. A Deep Neural Network (DNN) regressor is constructed and trained as the surrogate of the LT optimization process. The DNN-regressor outputs the spacecraft final mass with a prediction mean-relative error (MRE) of less than 1%. These two models are integrated into a grid search framework and enable efficient searches for LT journeys. For the given spacecraft configurations, 7% (1,684) of the 24,149 studied NEAs are LT round-trip accessible, and 95.4% of the LT accessible ones have minimum propellant mass fractions between 0.08 and 0.29. The identified LT accessible NEAs have inclinations less than 9 deg and eccentricities less than 0.4. Some asteroids, such as 2017 CF32, are found to be more accessible by the low-thrust propulsion option than the impulsive propulsion. The results of this study can be used as a reference for future low-thrust NEA mission target selection.