Fault Isolation of Reaction Wheels for Satellite Attitude Control

Abstract

A new hierarchical method for fault detection, isolation, and identification of nonlinear systems with applications for in-orbit closed-loop controlled satellites is proposed where fault detection employs unscented Kalman filter and adaptive thresholds. Fault isolation employs multiple model approach in conjunction with Bayes’ probability theorem and adaptive window, and fault identification employs dual state/parameter estimation using unscented Kalman filter. Results show up to 97% success rate under comprehensive Monte Carlo simulations for transient abrupt faults in satellite actuators.