Advancing Conjunction Analysis in SDA Through Hypothesis Reasoning and Analytical Fusion

Rapid advancements in commercial and governmental space missions have led to an unprecedented increase in satellite deployments, particularly in low Earth orbit (LEO), resulting in more complex and dynamic operational environments. There are several challenges related to satellite operations, one of them is conjunction analysis, which helps identify and manage space object proximity events and potential orbital collisions. Traditional conjunction analysis methods applying covariance-based techniques and probabilistic integration can provide accurate predictions, they often assume ideal or generic conditions and complete information about orbital uncertainties, where the analysis results are not conclusive in all cases. Given this context, in addition to the large volume of data related to space operations available and the increasing number of satellites in orbit, there is growing interest in using data-driven and machine learning methods to improve prediction accuracy and better support decision-making. This indicates a broader shift toward more intelligent and adaptive systems that can handle the growing amount of data and complexity in operations. To support this development, some recent work has looked into cognitive engineering applications integrating human input to boost situational awareness and decision-making. This paper proposes a method for hypothesis generation and evidence analysis applied to satellite conjunction analysis, contributing to the development of an advanced system for collision risk management to support satellite operators in collision avoidance strategies.