Low-thrust trajectories: For Deep Learning Model Construction

Published in Journals, 2021

This database stores randomly generated low-thrust(LT) trajectories between NEAs. This database can be used for training interesting machine learning models. For example, using both feasible LT samples and infeasible samples to build a DNN classification model to distinguish LT transfer feasibilities. This database is valuable – obtaining such database requires hundreds of thousands of cores hours of high-performance computing system.

Database Summary

ItemDescription
Number of feasible trajectories300,000
Number of feasible trajectories2,000,000
File Format.MAT
File Size3 GB
Precisiondouble-precision floating-point
Columnsexitflag, SPK1, SPK2, depart date, tof, spacecraft mass, PV1, PV2, Control Vectors

Statistical

 Time of flight (days)Initial Mass (KG)Lambert Delta-V (KM/S)
Min12610530.15
Max14603000116.27
Mean875.91830.545.00
Median883173241.65

DB3-stats1 DB3-stats2 DB3-stats3

The link will be released later. Contact: ruida.space@gmail.com