Volunteer Computing for a Global Energy-Regime Census of One Million Cislunar Orbits and a Deep Learning Surrogate for Rapid Stability Prediction

The cislunar space, governed by the circular restricted three-body problem (CR3BP) butsubject to additional perturbations in high-fidelity models, presents significant challenges for mission design due to its complex stability structure. Traditional numerical integration is computationally prohibitive for a systematic energy-regime census of millions of orbits. Here, we present a novel approach based on global volunteer computing via the BOINC platform to overcome this barrier. Using the public “Million Orbit” dataset from Lawrence Livermore National Laboratory (generated with a high-fidelity model including solar and planetary perturbations), we distributed the computation of time-resolved Jacobi constant sequences across thousands of volunteer devices, producing over 16 billion individual values. The resulting dataset is freely available. Analysis reveals that the majority of orbits are high-energy escapes (Region V), while a non-negligible fraction belong to the low-energy Region I, with a small number of intermediate cases. A single rare Region IV orbit (ID 754482) is identified and analysed. Furthermore, we develop a lightweight deep learning surrogate that predicts whether an orbit belongs to Region I using only the first K Jacobi constants. Our model combines an LSTM encoder with attention and an XGBoost classifier, achieving test AUC of 0.984 with K = 500 and 0.929 even with K = 10, outperforming a raw XGBoost baseline. This work demonstrates the transformative potential of volunteer computing for large-scale astrodynamics and provides an efficient machine learning tool for real-time orbit screening.