Multiple model filters applied to wind model estimation for a fixed wing UAV

The flight of unmanned aerial vehicles is often associated with model uncertainties, measurement noises, and environmental disturbances such as wind gust. To mitigate these challenges, the accurate estimation of states is vital. Moreover, the wind model and its parameters should also be estimated and compensated during the flight. In this paper, a multiple model filter is implemented for this purpose. To investigate the performance of the multiple model filter, three different models including constant wind, “1-cosine” model and wind shear model are considered. The multiple model filter utilizes three extended Kalman filter to simultaneously estimate the model of wind, the parameters of the model as well as the current states. Simulation results show that the multiple model filter provides good performance and the wind model is properly estimated. Moreover, small estimation errors, obtained from the simulations, prove the good performance of this approach in estimation of states, wind model, and its parameters.