A Novel ESPRIT Algorithm for Analysis of Low Frequency Oscillations in Power System

Abstract

With rapidly increasing load and randomly varying load scenario in a densely connected power system may lead to low-frequency oscillations which can further make a power system unstable. To keep the power system in stable operating limits identifying such dominant low-frequency oscillations is essential. Numerous signal processing techniques are used to detect low-frequency oscillations in a power system. Different techniques have their advantages and disadvantages. Estimation of signal parameters via rotational invariance technique (ESPRIT) is a signal processing technique by which low-frequency oscillations can be identified even under noisy conditions . In this paper, a novel ESPRIT algorithm for identifying low-frequency oscillations is proposed. This methodology is based on the ESPRIT algorithm, whose key concept is that real modes existing in oscillations emerge consistently regardless of the algorithm's order. The performance of this algorithm is verified through a set of synthetic signals generated in MATLAB with different noise levels and varying PMU rates and compared with traditional Prony algorithm. The performance of the proposed algorithm is also verified on real-time test data obtained from the Western Electricity Coordinating Council (WECC). This technique performs fairly well in finding low-frequency oscillation characteristics in the power grid.