Icing detection and identification for unmanned aerial vehicles using adaptive nested multiple models

A multiple-model approach for icing diagnosis and identification in small unmanned aerial vehicles is proposed. The accretion of ice layers on wings and control surfaces modifies the shape of the aircraft and, consequently, alters the performance and controllability of the vehicle. Pitot tubes might be blocked due to icing, providing errors in the airspeed measurements. In this paper, we propose a nested multiple-model adaptive estimation framework to detect and estimate icing using standard sensors only, ie, a pitot tube and an inertial measurement unit. The architecture of the estimation scheme is based on 2 different time scales, ie, one for the accretion of ice on aircraft surfaces and one for the accretion of ice on sensors, and consists of 2 nested adaptive observers, namely, outer and inner loops, respectively. The case study of a typical small unmanned aerial vehicle supports and validates the proposed theoretical results.