An Improved Unscented Kalman Filter for Interrupted and Drift Sensor Faults of Aircrafts

Reliable sensor signals are key factors in the flight control system. The development and application of various sensors not only improve flight performance and maneuvering capability but also bring challenges to the reliability and safety of the system. When sensor faults occur in nonlinear aircraft models, it is crucial to quickly recover system states. In this article, a new state estimation approach is presented using an interrupted-estimation unscented Kalman filter (IEUKF), which is integrated with maximum-likelihood estimations and Gaussian probability distribution. In general, aerial sensors involve complex fault conditions and a large number of strongly nonlinear systems. In spite of that, the proposed method can compute estimations robustly and achieve high precision under hard/soft and interrupted/permanent sensor faults. Compared with cutting-edge methods, the IEUKF can be applied to featureless interrupted faults without redundant sensors, foregone-fault hypotheses, or slow-varying fault assumptions. With a nonlinear F16 Simulink model as the example in simulations, the results illustrate the advantages of the proposed method.