The Augmented Unscented H-infinity Transform with H-infinity Filtering for Effective Wind Speed Estimation in Wind Turbines

Wind energy is a fast-growing industry in Canada and worldwide. As wind turbine size and capacity increase, control systems become exceedingly important in order to maximize the efficiency of the power output and to reduce loads to extend their longevity. Effective wind speed (EWS) is not easy to measure because wind speed varies in time and space across the rotor area of the turbine, nor is it a real physical signal. This research aims to provide better knowledge of the input wind speed and to design a turbine controller based on this normally unknown input. First, nonlinear robust methods of state estimation were considered in order to deal with the nonlinearities present in the wind turbine model and the large exogenous disturbance of wind speed. Specifically the unscented H-infinity filter was used for robustness and accuracy. This unscented H-infinity scheme was then adopted in combination with a data fusion technique to estimate effective wind speed. This technique utilized high frequency data from the anemometer and treated the turbine as a sensor of low frequency data. These signals were fused to create an EWS estimate. The methods are thus shown to be effective in wind speed estimation for two sizes of turbines.