Space-based Maneuver Detection and Characterization using Multiple Model Adaptive Estimation

An increasingly congested space environment requires real-time and dynamic space situational awareness (SSA) on both domestic and foreign space objects in Earth orbits. Current statistical orbit determination (SOD) techniques are able to estimate and track trajectories for cooperative spacecraft. However, a non-cooperative spacecraft performing unknown maneuvers at unknown times can lead to unexpected changes in the underlying dynamics of classical filtering techniques. Adaptive estimation techniques can be utilized to build a bank of recursive estimators with different hypotheses on a system\u27s dynamics. The current study assesses the use of a multiple model adaptive estimation (MMAE) technique for detecting and characterizing noncooperative spacecraft maneuvers using space-based sensors for spacecraft in close proximity. A series of classical and variable state multiple model frameworks are implemented, tested, and analyzed through maneuver detection scenarios using relative spacecraft orbit dynamics. Variable levels of noise, data availability, and target thrust profiles are used to demonstrate and quantify the performance of the MMAE algorithm using Monte Carlo methods. The current research demonstrates that adaptive estimation techniques are able to handle unknown changes in the dynamics while keeping comparable errors with respect to other classical estimation methods