Aircraft Inflight Icing Detection Based on Statistical Decision Theory

Authors:
Open Access
Publisher website
Google Scholar
Abstract
Aircraft inflight icing detection is very important for aviation flight safety. The ice alarm signal given by detectors can help the pilot to actuate the ice protection system. The application of a Neyman Pearson (NP) based statistical decision approach in icing detection is discussed in this paper. First, the flight dynamics for icing aircraft is modelled, the dynamic behaviour of an airbus influenced by ice accreting on the leading edge of wings is analysed. Then the residuals which are directly affected by the aerodynamic parameters are presented and their generation way is given. The Generalized Likelihood Ratio Test (GLRT) method is utilized in detecting the changes of residuals inflight, its performance is assessed by a flight scenario with wing ice. The results show that the fault of wing icing can be detected by every residual under different window sizes. By comparing the results of the residuals, it can be concluded that the small window size will degrade the detection performance with the longer alarming time delay and the lower probability of detection; appropriately increasing the window size can reduce the alarming time delay; diagnosing by residual of normal overload is more reliable than by the other two residuals.

This publication has 19 references indexed in Scilit: