Abstract:

In this paper, we investigate the fundamental frequency and phase estimation problem in a balanced three-phase power system with harmonic distortion. An estimation algorithm that can rapidly track the fast-changing frequency and phase by using quarter cycle samples are proposed. Specifically, the data model in the three-phase power system is first converted to the noise-corrupted single phase harmonic signal model by using the Clarke transformation. A new weighted least squares (WLS) parameter estimator, which refines the initial estimates from the standard estimation techniques, is computed by utilizing the harmonic structure of the signal. Since the initial estimates become unreliable with limited samples, we proposed an iterative algorithm to polish the initial estimates for WLS. Numerical results show that the proposed estimator outperforms the conventional estimators, especially in a data-limited case.