Tim Glinski, LSAS Tec; Drew Latta, LSAS Tec; Andrew Allen, LSAS Tec
Keywords: UCT, SDA, Debris
Abstract:
LSAS Tec has, through automation of ANSYS’s ODTK and integration with proprietary algorithms, succeeded in developing an Un-correlated Track (UCT) processor that can distinguish between multiple orbiting objects in close proximity using a filter and smoother for processing.
The final stage of UCT processing is often done by calculating an ephemeris file that best matches the data using a Batched Least Squares solution. These results are then scored based on RMS values. However, when multiple objects are in close proximity, data sets with multiple objects – often referred to as cross-tags – can be solved for as a single object with minimal RMS, giving the faulty impression that the system has arrived at a quality solution. Without proper detection of these issues, predictions can diverge, and reacquisition of objects is difficult. Currently, these issues are then solved via manual operations by highly skilled orbit analysts.
Inconsistencies between different groups of data can be detected by applying the McReynold’s consistency test on a highly refined orbit. This allows automatic detection of cross-tags and thus proper evaluation, allowing for higher confidence processing of UCT data. This, however, can only be achieved from an ephemeris with high-precision covariance, which requires high-fidelity models that consider all sources of process noise in a dynamic model. ODTK’s Extended Kalman Filter and Smoother does this, allowing for this post-analysis to be conducted.
Additionally, a methodology for distinguishing between a dataset with multiple objects as opposed to one with noisy observations must be developed. LSAS-Tec’s solution applies statistical tests to confirm consistency in data subsets before looking for consistency between the data and a full satellite state estimation model. This separation and staging of model analysis allows for the detection of outlier data, cross-tags, and satellite model errors to all be done in isolation. This is essential for automated corrections to be conducted with confidence.
The LSAS-Tec UCT solution is split into a first stage where data is grouped into nodes and outliers are identified. This is followed by an automatic IOD and Batched Least Squares run that implements a modified version of a RASTER algorithm. Finally, the orbit solution and overall model is further refined and force models are dynamically tuned as accurate uncertainty models are created. Steps are taken at each stage to prune out bad solutions, determine dynamic and statistical consistency and identify any overall solution divergence. Afterwards, better fitting data groups are searched for, and the same tests and refinements are continuously performed.
This methodology has proven to produce high quality ephemeris solutions comparable to what a well-trained orbit analysist could produce.
This paper shows the results of applying these methods to process UCT data from a variety of sources, including observations both on objects that are thought to be debris as well as actively moving satellites. Additionally, this paper seeks to compare this methodology and its results to that of a Batch Least Squares solution.
Date of Conference: September 16-19, 2025
Track: Astrodynamics