This paper proposes a novel Kalman filtering based dynamic state estimation method, which addresses cases of models with a nonlinear unknown input, and it is suitable for wind turbine model state estimation. Given the complexity characterizing modern power networks, dynamic state estimation techniques applied on renewable energy based generators, such as wind turbines, enhance operators' awareness of the components comprising modern power networks. In this context, the method developed here is implemented on a doubly-fed induction generator based wind turbine, under unknown wind velocity conditions, as opposed to similar studies so far, where all model inputs are considered to be known, and this does not always reflect the reality. The proposed technique is derivative-free and it relies on the formulation of the nonlinear output measurement equations as power series. The effectiveness of the suggested algorithm is tested on a modified version of the IEEE benchmark 68-bus, 16-machine system.