SpaceKG: Towards exploiting Knowledge Graphs in space systems

Abstract

Space systems consist of a vast quantity of deeply interconnected and time-evolving elements. This complexity has been steadily increasing in recent years, prompting the development of intelligent tools capable of managing information in such a scenario and transforming it into actionable knowledge. This paper aims to combine deductive artificial intelligence with space engineering, integrating space project activities with Knowledge Graphs’ semantics and operational dynamics. We propose SpaceKG, a real-time, data-driven, dynamically evolving cognitive digital twin that enables digital continuity throughout the life cycle and across the disciplines of space systems. While traditional Model Based Systems Engineering approaches focus on Knowledge Representation, they struggle with rapid adaptation to change. In contrast, our solution offers a dynamic framework to deal with faster interactions with domain experts and evolving requirements. We showcase and validate the effectiveness of the proposed approach in a specific case study for detecting failure events in the Space Shuttle Main Engine. To this aim, we leverage Vadalog, a high-performance deductive reasoning language that provides full transparency and explainability over the portion of the space system more comprehensively and flexibly than existing approaches.