An ML-Driven Mission Planning Pipeline for SmallSat Radiation Mitigation

As the democratization of space increases the number of amateur and university-led SmallSat missions, there is a growing need for accessible mission planning tools that simplify complex orbital mechanics and the impact of space weather on spacecraft. A critical parameter for mission longevity is the Earth’s magnetotail; a region on the night-side of the Earth where the magnetosphere is stretched by solar wind, creating a “shielded” volume against solar particle events (SPEs). This shielding is vital for science-centric missions, as it provides “radiation-free zones” required for low-noise instrument calibration, error-free data storage, and the protection of sensitive payload sensors from high-energy proton displacement damage.
This research introduces a pipeline designed to predict these shielded zones throughout a satellite’s orbit. The methodology utilizes a Long Short-Term Memory (LSTM) network to forecast magnetotail boundaries and dynamics. Unlike traditional static models, the LSTM architecture captures the temporal dependencies and time-lagged responses of the magnetotail to fluctuating solar wind conditions. The model ingests spacecraft Two-Line Element (TLE) data integrated with real-time solar wind indices to output highly accurate entry and exit timestamps.
This pipeline/ software allows SmallSat operators to autonomously schedule high-risk operations (such as memory-intensive data processing or the deployment of sensitive components) during periods of maximum magnetic shielding. The pipeline is designed as an open-source tool to bridge the gap between high-level magnetospheric physics and practical SmallSat mission operations, providing amateur scientists with enterprise-grade environmental intelligence.