Charles J. Wetterer, KBR; C. Channing Chow II, Cloudstone Innovations LLC; Jason Baldwin, Complex Futures LLC; Micah Dilley, KBR; Keric Hill, KBR; Helen Montell-Weiland, KBR; Paul Billings, KBR; Christopher Craft, KBR; James Frith, AFRL/RVSW
Keywords: Cislunar, Maneuver Detection, Passive RF
Abstract:
Orbit determination and tracking of objects in cislunar space is more challenging in part because of the added complexity of the orbital dynamics due to the influence of the Moons gravity. Despite this, in a previous paper we showed that the Unscented Kalman Filter (UKF) successfully converged to a solution in most situations when using an appropriate dynamics model to propagate the state. We employ a KBR capability known as the Infinity Filter Framework (IFF) to build the estimation filter together with NASAs General Mission Analysis Tool (GMAT) as the dynamics model. The simulated orbital states used in the analysis are generated using a different, yet comparable, high-fidelity dynamics model. In this paper, we apply the UKF in an Interacting Multiple Model (IMM) approach to identify maneuvers in simulated data of a differentially corrected halo orbit using passive RF observations. Passive RF observations were chosen to allow consistent daily observations during the day or night and to legitimize the assumption that the observations can be associated with the object despite the maneuver. We investigate maneuvers from the L1 periodic Halo family (i.e., H1) with a range of delta-V magnitudes (9 values between 0.1 m/s and 298.1 m/s) and directions (radial, in-track and cross-track). This explores both extremes of maneuver detection. For small velocities, the maneuver may be too small to trigger detection while for large velocities, the filter may diverge due to the larger uncertainties and non-linear dynamics. Observations are time difference of arrival (TDOA) and frequency difference of arrival (FDOA) values from a 3-site hypothetical ground-based passive RF array. The three baselines yield 6 total measurements (3 TDOAs and 3 FDOAs) for each time step chosen to be a 300 second cadence for this study. With a single Earth-based passive RF array, there are daily data gaps when the object is below the horizon as viewed by any one of the three sensor sites. To explore various detection possibilities, the maneuvers take place at 4 different epochs corresponding to the beginning of daily observations, midway through the daily observations, at the start of the data gap between observations, and midway through the data gap. Finally, the maneuver model in the IMM uses four different isotropic process noise values (0.1 to 100 m/s) to not only detect whether a maneuver takes place or not, but to also classify the magnitude of the maneuver. For a selection of maneuvers, multiple maneuver models with direction-specific process noise values are employed to demonstrate directional classification of the maneuver as well. We show how the IMM filters behave when processing the different cislunar passive RF observation sets and the effectiveness and limits to cislunar maneuver detection and classification.
Date of Conference: September 27-20, 2022
Track: Cislunar SSA