Autonomous Fault Diagnosis and Adaptive Multi-Source Fusion for Robust Spacecraft Navigation

Spacecraft navigation systems are critical for mission success, yet they face significant challenges from sensor faults induced by the harsh space environment. Specifically, Global Navigation Satellite System (GNSS) signals are susceptible to outages and interference, while Star Trackers can suffer from occlusions and miscalculations, posing a severe threat to navigation integrity. In this paper, we propose a robust and innovative architecture that integrates an autonomous fault diagnosis module with an adaptive multi-source sensor fusion framework to enhance the reliability and accuracy of spacecraft navigation. The proposed framework employs a model-based Fault Detection and Identification (FDI) module that uses statistical analysis of measurement residuals to identify failure modes. The discrete output of the FDI module is then used to dynamically modulate the information contribution of each sensor’s sub-filter within a multi-branch Federated Kalman Filter (FKF), enabling a fault-tolerant "soft-switching" response via a weighted information fusion strategy. Simulation experiments demonstrate that the proposed system effectively detects a comprehensive of GNSS and Star Tracker anomalies and sustains sub-meter positioning accuracy under these fault conditions These results confirm that our method offers a robust and high-integrity solution for autonomous spacecraft navigation in degraded sensor environments.