Ambient Data-Based Online Electromechanical Mode Estimation by Error–Feedback Lattice RLS Filter.

This paper proposes a novel error-feedback lattice recursive least-squares (EF-LRLS) filter for online estimation of power system oscillatory modes. The EF-LRLS filter is applied to ambient data provided by phasor measurement units to identify the autoregressive (AR) model parameters. This filter has a modular structure; accordingly, if the length of the filter equals $N$ , it identifies AR(1) to ${\rm{AR(}}N)$ models concurrently. In the proposed method, removing very low and high frequencies and resampling steps are fulfilled in an online fashion. This adaptive filter has less computational complexity than standard RLS filter, making it an appropriate choice for online system identification. The proposed algorithm is tested using simulated ambient data of the 16-machine, 68-bus test system, as well as real measurement data of the WSCC system breakup on Aug. 10, 1996. Performance of the proposed algorithm is examined under different conditions with different amplitudes of variation and signal-to-noise ratio. The results prove accuracy, robustness, and effectiveness of the proposed method. [ABSTRACT FROM AUTHOR]