Non-deterministic wind observation from wind turbine loads

Abstract

In this work, the wind sensing technology that exploits the turbine rotor as an anemometer is further developed into a non-deterministic formulation. First, an inflow-turbine response map is identified, which relates out and in-plane blade root bending moments to both vertical and horizontal shears and misalignments. Then, this linear model is used with and without a Kalman filter to estimate online the wind field at the rotor disk once blade loads are measured. A comprehensive simulation study, including different wind speeds and turbulence intensity levels, was performed to evaluate the accuracy of the new non-deterministic formulation. The results show that introducing a Kalman filter in the estimation process allows for a significant improvement in the angle estimates with respect to the deterministic formulation, with no considerable additional computational cost.