An LSTM-based Maneuver Detection Algorithm from Satellites Pattern of Life

Near Earth Environment is swiftly turning into an overpopulated operational space, mainly due to increased commercial missions and service-aimed constellations. As a consequence, the development of an efficient Space Traffic management infrastructure is progressively becoming a mandatory requirement. In this framework, Space Surveillance and Tracking programs play a key role by taking care of the entire measurement processing pipeline and maintaining Resident Space Object catalogs by updating orbital data for each tracked target. Collecting a vast quantity of structured data represents the perfect use-case for data-driven techniques to mine for hidden patterns and features within them. This work shows how a Long-Short-Term-Memory Neural Network, specialized in time sequences analysis, can take advantage of an operational object's Pattern of Life, consisting of its state and maneuvering history, and perform maneuver detection on new incoming orbital parameter sequences. These data prove fundamental in progressively labeling a target orbit evolution, characterizing its operational life, and detecting mission phase changes. They also help in providing a deeper context to an operator performing any of the following tracking-related activity, adding background information retrieved from the effective processing of a target's history.

[1]  Cláudia Soares,et al.  Machine Learning in Orbit Estimation: a Survey , 2022, Acta Astronautica.

[2]  M. Massari,et al.  Real-time space object tracklet extraction from telescope survey images with machine learning , 2022, Astrodynamics.

[3]  Zoltán Gál,et al.  Recurrent neural network variants based model for Cassini-Huygens spacecraft trajectory modifications recognition , 2022, Neural Computing and Applications.

[4]  R. Jaimes,et al.  Satellite Maneuver Detection Using Machine Learning and Neural Network MethodsBehaviors , 2022, 2022 IEEE Aerospace Conference (AERO).

[5]  Manuel Sanjurjo-Rivo,et al.  Automatic maneuver detection and tracking of space objects in optical survey scenarios based on stochastic hybrid systems formulation , 2021, Advances in Space Research.

[6]  Xiao Pan,et al.  Mining Two-Line Element Data to Detect Orbital Maneuver for Satellite , 2019, IEEE Access.

[7]  Rohit Mital,et al.  A Machine Learning Approach to Modeling Satellite Behavior , 2019, 2019 IEEE International Conference on Space Mission Challenges for Information Technology (SMC-IT).

[8]  Alex Sherstinsky,et al.  Fundamentals of Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) Network , 2018, Physica D: Nonlinear Phenomena.

[9]  Tim Flohrer,et al.  Correlation of Observations and Orbit Recovery Considering Maneuvers , 2017 .

[10]  H. Krag,et al.  Two-Line-Elements-Based Maneuver Detection Methods for Satellites in Low Earth Orbit , 2014 .

[11]  Russell P. Patera,et al.  Space Event Detection Method , 2006 .

[12]  Sepp Hochreiter,et al.  The Vanishing Gradient Problem During Learning Recurrent Neural Nets and Problem Solutions , 1998, Int. J. Uncertain. Fuzziness Knowl. Based Syst..

[13]  S. Hochreiter,et al.  Long Short-Term Memory , 1997, Neural Computation.

[14]  R. Linares,et al.  Geosynchronous satellite maneuver classification via supervised machine learning , 2021 .

[15]  Phil DiBona.  Machine learning for RSO maneuver classification and orbital pattern prediction , 2019 .

[16]  Roberto Furfaro,et al.  Space Objects Classification via Light-Curve Measurements: Deep Convolutional Neural Networks and Model-based Transfer Learning , 2018 .

[17]  Charlotte Shabarekh,et al.  A Novel Method for Satellite Maneuver Prediction , 2016 .

[18]  Tom Kelecy,et al.  Satellite Maneuver Detection Using Two-line Element ( TLE ) Data , 2007 .