Accurate Estimation of Effective Wind Speed for Wind Turbine Control Using Linear and Nonlinear Kalman Filters

Wind speed measurement depends on the efficiency of the anemometer sensors. In fact, the anemometer wind speed sensors, mounted on top of the wind turbine nacelle, give the wind speed measurements to the controller devices. These measurements obtained from the anemometers represent the turbulent wind speed and cannot represent the wind speed upstream of the rotor blades. For that, Effective Wind Speed (EWS) will be estimated by applying linear and nonlinear Kalman Filter methods to Wind Turbines System (WTS). In this study, the variable wind speed is computed numerically based on the estimation values obtained from the linear KF, while it is directly estimated from the nonlinear KF. Furthermore, Backward Euler Approximation is used to perform the discretization of the continuous-time state space model of WTS based on KF estimators. In this paper, the efficiency of each filter is investigated on a wind turbine using a two-mass drive train interconnected by a common spring and damper. The comparative studies show that the two estimators have high accuracy estimates at the steady-state with greater than 95% similarity between the estimated and measured wind speeds. However, the nonlinear KF has better performance than linear KF at the transient-state due to its very small time response. These results achieved are justified by computing the covariance and the Correlation Coefficient (CC) between the measured and estimated wind speeds for each 5 ms. The covariance curves obtained are positive, while the CC curves are close to one at the steady-state for both estimators.