Fault detection and isolation based on UKFs for a novel ducted fan UAV

摘要

This paper presents an online nonlinear actuator fault detection and isolation system, aiming at monitoring monitor the health of a novel ducted fan UAV. In order to cover wider flight envelope, an unscented multiple model adaptive estimation method is introduced as an extension of the EKF-based method. The proposed method provides a bank of UKFs running in parallel, each of which is responsible for completely monitoring corresponding actuator's health. Simulation results shows that locked-in-place and floating faults of the control vanes and failures of the auxiliary ducted fans can be handled with small ambiguity or false detection, rapid response and low computational load, indicating a more efficient operation than EKF-based method. Then, the robustness of the fault detection system is further enhanced by the usage of auxiliary excitation signals.