Data-driven methods for the analysis of wind turbine yaw control optimization

Multi-megawatt wind turbines are nowadays a mature technology, and therefore, there is considerable scientific and industrial attention to the opportunity of further improving the efficiency of wind kinetic energy conversion into electricity. One of the major developments in this field of research regards the optimization of wind turbine control. This work deals with a test case of yaw control optimization on a 2-MW wind turbine sited in Italy. The objective of the work is to compute the performance improvement provided by the upgrade after some months of operation. This has been accomplished through the formulation of an appropriate model for the power of the wind turbine of interest and the analysis of the residuals between model estimates and measurements before and after the upgrade. In this work, a general procedure for selecting a robust multivariate linear model is adopted, and the resulting model, employing as input variables several operational variables from the nearby wind turbines in the farm, is used for quantifying the performance improvement. The estimate is that this upgrade provides a 0.8% improvement of the annual energy production.