A Robust Unscented Kalman Filter for Intermittent and Featureless Aircraft Sensor Faults

Fault tolerance is crucial to reliable state estimations in aircrafts. When time-varying sensor faults are coupled with nonlinear system models, it is non-trivial to robustly recover system states. In this paper, based on an intermittent-measurement unscented Kalman filter (IMUKF), we propose a new state-estimation method by combining the IMUKF with maximum-likelihood estimation and Gaussian mixture reduction. For nonlinear systems under time-varying and featureless sensor faults, the proposed method performs robust estimation without past error history or explicit fault detection, achieving significantly enhanced accuracy under severe conditions. With a nonlinear twin-spool jet engine as the example in simulations, the effectiveness of the proposed method is verified.