An aerodynamic model of a 4.6% scaled version of the Flying-V is estimated
using flight data gathered in eight test flights. The two-step method used
consists of first, the state estimation using an Iterated Extended Kalman
Filter(IEKF), and second the polynomial model structure is determined
using a stepwise regression algorithm. Using the state estimate a stepwise
regression performed a model structure estimation. The longitudinal
models generally also perform better, with an average RRMSE of 13.2%
compared to 13.4% for the lateral models. In addition, more higher order
terms were needed for the lateral models. This indicates that more
informative data is needed to model the lateral system behaviour. For
CX it is found that the current thrust model is not adequate. The model
for the pitching moment gave a large deviation from the model that was
found during wind tunnel tests. In addition, for Cn a clear relationship
with β could not be found. Both of this is likely due to the relatively low
level of excitations present in the data set.