ROBUST RTSS-BASED ESPRIT METHOD FOR LOW FREQUENCY OSCILLATIONS ANALYSIS

Low frequency oscillations (LFOs) threaten the stability of power systems. The estimation of signal parameters via rotational invariant techniques can analyse LFOs with high accuracy only when the model order of the analysed signal is known. This paper proposes a novel model order estimation method for modal analysis of LFOs. The method first builds a singular energy spectrum to inspect whether the measurement data is being polluted by complex interferences (e.g., impulsive noises). Then, a tailored Rauch-Tung-Striebel smoother is utilized to alleviate the impact of complex interferences. Afterwards, the mean value of the singular energies is adopted to determine a rough estimation of the model order of dominant modes in LFOs. Finally, the reconstruction quality indicator of the reconstructed signal is introduced for detecting and correcting overestimation and fake modes. The proposed solution is experimentally evaluated via simulations and field measurement data obtained from the phasor measurement unit installed at a generating station in North America. Results show that the method is accurate, robust, and suitable for field applications.