Flying quality assessment with EKF-derived aerodynamic derivatives from weather-hazardous FDR data

Weather hazard has been one of major sources in flight accidents and often leads to injuries. According to a flight accident report of the U.S. Federal Aviation Administration (FAA), close to 20% of accidents were caused by weather hazard [FAA, 2009]. In this study, the FDR data from a flight accident affected by cross wind have been processed. Firstly, the Extended Kalman Filter (EKF) was applied to perform the kinematics compatibility check and to estimate unrecorded parameters, such as the 3-dimensional wind field, the sideslip angle and the rotation rates in the body axis. Secondly, the radial based function neural network (RBFNN) method was used to compute both longitudinal and lateral aerodynamic derivatives.