Stochastic wind modeling and estimation for unmanned aircraft systems

This paper presents a sensor fusion technique for wind estimation using two different stochastic models for wind speed: random walk and Gauss-Markov. The parameters for these models are formally derived from experimentally collected weather station data using an Allan deviation approach. Using these parameters, the two different stochastic wind models are then implemented within a nonlinear Kalman filtering approach to wind estimation and compared using two sets of unmanned aircraft flight data. This work showed that even though the Gauss-Markov model can more closely model the distribution of wind speed, only small differences are noted when comparing to the commonly implemented random walk model.