The use of Wavelets for Determining Wing Flexure in Airborne GPS Multi-Antenna Attitude Determination Systems

In the quest to obtain a precise set of aircraft attitude parameters the modeling of wing flexure can become an important issue, especially when short baselines are being used. Previous techniques for the estimation of wing flexure have relied on a constrained Least Squares process to determine the displacement at each epoch before the estimation of the attitude parameters. As will be shown in this paper, there are more effective techniques available for the estimation and removal of wing flexure. If the variation in the height of the antenna is viewed as a signal, digital signal processing techniques can be implemented, such as Discrete Fast Fourier Transforms (DFFT), to spectrally analyze the signal. This information can then be used to construct an equiripple FIR filter to isolate/remove the wing flexure component from the original signal. Another approach is to use a wavelet technique. The coefficients obtained from a wavelet transformation can be manipulated to filter the noise from the wing flexure signal. Only the coefficients corresponding to the wing flexure are used in the reconstruction process thus effectively filtering the unwanted noise from the original signal. This paper will assess the constrained Least Squares solution compared with the DFFT and wavelet techniques through a series of simulations and experiments. The results obtained highlight the superior ability of wavelet techniques to denoise stationary and in particular nonstationary signals.