PHM for Spacecraft Propulsion Systems: Developing Resilient Models for Real-World Challenges

Abstract

This paper extends the research presented at the Prognostics and Health Management (PHM) Asia-Pacific 2023 Conference Data Challenge, focusing on a more pragmatic approach to spacecraft propulsion system health assessment. While the previous competition saw a variety of solutions, they predominantly relied on the assumption of highly stable initial hydraulic conditions – an idealization seldom met in real-world scenarios. In practical settings, factors such as operational noise, recent operational states, and ambient environmental conditions significantly disrupt this stability, rendering such solutions less feasible. Addressing this gap, our current study introduces a novel diagnostic model capable of valve faults without depending on the initial stable state of hydraulics. This approach marks a significant shift from our previous methodology, which primarily utilized similarity measures and physics-inspired features to classify health states and identify solenoid valve faults in spacecraft propulsion systems. The proposed model in this paper is validated against a diverse set of conditions, emphasizing its robustness and applicability in fluctuating real-world scenarios. Our findings demonstrate that the new model not only effectively diagnoses system health under varied and less controlled conditions but also enhances the practicality of spacecraft health management, offering a more adaptable solution in the face of operational uncertainties.