Alexander Ryan, Industrial Sciences Group; Angus Leung, Industrial Sciences Group; David Shteinman, Industrial Sciences Group; Mark Yeo, Industrial Sciences Group; Matthew Hejduk, NASA Conjunction Assessment Risk Analysis (CARA)
Keywords: Space Object Conjunction Assessment (CA);Space Traffic Management; Big Data analytics
Abstract:
As the quantity of orbital debris continues to grow, so too does the rate of conjunction messages that suggest possible collisions between high value payloads and debris. The abundance of these conjunctions, and eventual misses, has led to a culture of ignored warnings, and an increase in satellite operation costs as a result of maneuver planning. The loss of “trust” in conjunction warnings is due to the poorly characterised evolution in probability of collision (Pc) as time approaches the time of closest approach (TCA) between two objects, as well as the interpretation of Pc in the context of maneuver planning. To address these problems, and in collaboration with the NASA Conjunction Assessment Risk Analysis (CARA) program , the Industrial Sciences Group has developed a novel Decision Support System (DSS) to assist satellite operations in conjunction assessment and manoeuvre planning by using a meaningful and intuitive metric, based on the physical dynamics of conjunctions. Pc computation based on current CDM data is currently the primary tool used for satellite conjunction monitoring and planning and does not take into account future changes to Pc as time progresses.
The objective of the DSS is to provide satellite operators with a rigorous recommended course of action up to two days ahead of a maneuver commitment point, derived from Pc computations, following the detection of a conjunction. The DSS is comprised of three distinct components:
Prediction of the future statistical evolution of position covariance of both objects
Prediction of the potential change in miss distance between both objects
Interpretation of predicted trends into actionable maneuver decisions using a weighted urgency metrics
The first and second components use a series of thirteen Gaussian process regression (GPR) models, six models for each object covariance and one for miss distance volatility. This method was selected as it allows for the quantification of uncertainties in each Pc computation, and no restrictive assumption about the data is imposed. It was found that by analysing historical conjunction events, the trend in the event Pc over time can be predicted using GPR models when compared to test data. The GPR framework was evaluated and trained using the conjunction data message (CDM) database provided by CARA.
The third component is an interpretation of the predictions produced by the first two components, as well as other meaningful conjunction information such as current Pc and time to maneuver commitment point, quantified as time variable metrics in the range [0, 1]. Each of these metrics was combined using a weighted average to produce a single quantified urgency metric. The urgency metric was sorted into distinct maneuver recommendation bins including event dismissal, event monitoring, preparatory mitigation planning, substantial mitigation planning, and maneuver execution. The DSS comprised 25 adjustable parameters associated with each metric function, metric weightings, and recommendation bins. To effectively set and tune these parameters, ISG collaborated with the NASA CARA team to ensure that outputs reflected an intuitive, interpretable representation of conjunction information and provide advisory actions that correspond with those that would be taken by experienced NASA CA operators.
The produced urgency value was demonstrated to be both a more advanced filtering method for mission planning, as well as providing information for the specific type of planning required 2 days prior to TCA. ISG performed an analysis of 2,318 conjunction events (where Pc > 1e-8 at some point during the event) as part of the evaluation process. Of those events, the tool successfully predicted 76% 2 days ahead of time when the Pc is above 1e-4, and 82% 2 days ahead of time when the Pc is below 1e-7.
The tool is unique in the field of space traffic management as it combines physical and statistical models to provide operators with a rigorous recommended course of action well ahead of a maneuver commitment point. The tool is undergoing further refinement at CARA and by ISG to prepare for commercial deployment in later 2022.
Date of Conference: September 27-20, 2022
Track: Conjunction/RPO