Robust statistic dual Kalman filter for wind generator identification in presence of outliers

Abstract

Among the technologies for clean energy generation, wind turbines are one of the most advanced. Due to stochastic behavior of the wind, a proper control action must be exerted over these turbines in order to get a better use of the output power. A propel control action usually implies the knowledge of the system dynamics, and that is why the identification stage is also crucial. In this paper, the authors propose a robust statistics dual Kalman filter for identification of a wind generator. Although dual Kalman filter (DKF) has been used in system identification, its performance heavily depends on the absence of outliers in the measured data. However, outliers might be easily produced during the data acquisition stage. In this study, the authors show how by combining the DKF and the robust statistic Kalman filter, the problem of outliers in systems identification can be avoided. The method was implemented in Matlab R2009a®. Results for the identification of a wind generator working online are shown and compared to the performance of dual Kalman filter.