Strong Tracking Unscented Kalman Filter for Identification of Inflight Icing

Abstract

Aircraft icing degrades aerodynamic performance and poses safety risks, especially under nonlinear and uncertain conditions. In order to identify inflight icing in real time, this work proposes a Strong Tracking Unscented Kalman Filter (STUKF) which integrates the Unscented Kalman Filter (UKF) with an adaptive fading factor from strong tracking theory. The proposed STUKF improves robustness and responsiveness without requiring Jacobian matrices. A nonlinear airplane model with six degrees of freedom is used, with icing effects represented by a time-varying severity parameter estimated through state augmentation. Simulations are conducted under varying turbulence intensities and icing scenarios, including both gradual ice accretion and sudden ice shedding. When it comes to tracking speed and precision, the results demonstrate that STUKF performs better than the normal UKF. Notably, STUKF identifies sudden drops in icing severity within 12 s even under strong disturbances. STUKF also maintains stable performance across light to heavy turbulence levels. These findings demonstrate the effectiveness of STUKF for timely and reliable icing diagnosis, supporting its potential integration into smart icing protection systems or adaptive flight control strategies.