Data-driven Health Management System for Multi-Spacecraft Formation Flying

Multi-spacecraft missions involving a fleet of small satellites are gaining significant interest due to low-cost, versatility through reconfiguration, and robustness to failure via redundancy. However, having multiple spacecraft increases complexity of the system and likelihood of fault occurrence due to large number of components involved. A health management system (HMS) that can effectively detect and identify faults within a complex system is required. In this work, we propose a novel data-driven HMS for multi-spacecraft system that is inspired by artificial immune system (AIS) in biology. The developed HMS utilizes notion of self and non-self to distinguish nominal conditions from off-nominal states due to fault. In the process, antibodies (detectors) are generated from dataset of nominal flight data and support vector machine algorithms is employed for classification in the high-dimensional feature space. The proposed HMS is applied to a design reference mission that involves a fleet of spacecraft performing on-orbit inspection in low Earth orbit. The performance and capabilities of the architecture is validated through numerical simulations where spacecraft in the network are subjected to various faults. The result show promise of AIS-based HMS in effectively detecting and identifying faults for multi-spacecraft systems.