Abstract
Series compensation grids connected with type-3 wind turbine generator (WTG)-based wind farms have suffered numerous subsynchronous oscillation (SSO) events worldwide. For early alerting of SSO and effective development of protection and control strategies, it is critical to monitor and identify SSO accurately and quickly. Ambient data is continuously available, which is useful for online monitoring. This paper proposes an ambient data-driven SSO online monitoring method based on the Kalman filter (KF) combined with the multi-model partitioning filter (MMPF). The KF is utilized to fit the measured ambient data with an auto regressive (AR) model. Then, the damping factor (or damping ratio) and frequency in the SSO mode can be acquired by solving the roots of the characteristic polynomial corresponding to the AR model. Moreover, the MMPF is an effective model order selection method applied to the KF for better identification. The performance of the MMPF-KF method is demonstrated by simulations and real-time experiments. The results of case studies validate the effectiveness of the proposed method under various conditions.