A unified modelling and operational framework for fault detection, identification, and recovery in autonomous spacecrafts

Recent studies have focused on spacecraft autonomy. The traditional approach for FDIR (Fault Detection, Identification, Recovery) consists of the Run-Time observation of the operational status to detect faults; the initiation of recovery actions uses static Pre-Compiled tables. This approach is purely reactive, puts the spacecraft into a safe configuration, and transfers control to the ground. ARPHA is an FDIR engine based on probabilistic models. ARPHA integrates a High-Level, a Low-Level, and an Inference-Oriented formalism (DFT, DBN, JT, respectively). The Off-Board process of ARPHA consists of the DFT construction by reliability engineers, the automatic transformation into DBN, the manual enrichment of the DBN, and the JT automatic generation. The JT is the On-Board model undergoing analysis conditioned by sensor and plan data. The goal is the current and future state evaluation and the choice of the most suitable recovery policies according to their future effects without the assistance of the ground control.