LAUSR

Closed loop control of the rotor angular speed of variable speed wind turbines in below rated operation area has been of particular research interest as a means of improving their power conversion efficiency. Quite promising approaches are based on tracking of the optimum rotor angular speed reference, which depends on the effective wind speed that can be calculated from the estimated aerodynamic torque of the rotor. For this purpose, Kalman filtering has been proposed, due to the stochastic dynamics involved. However, there are challenges on the proper selection of the process noise covariance in the Kalman filter, due to the nonstationary wind statistics. In this paper, an effective solution to overcome these challenges is proposed, using the multiple model adaptive estimation method. As shown from hardware-in-loop simulation results with real wind turbine and wind data, this method effectively tackles the uncertainty in the process noise covariance and exhibits remarkable estimation accuracy under all wind conditions.