Distributed Parametric Identification of Low Frequency Oscillatory Modes in Multiple PMU

Abstract

This paper analyses performance in the estimation of low frequency oscillatory modes utilizing Prony's method of parametric identification applied to multiple signals (multiProny) in distributed form. The analyzed methodology applies the multiProny method by area and integrates local modal estimates in a consensus optimization process that determines a global solution for the entire system. In each area, multiple measurements are processed corresponding to ringdown data, recorded in case of contingencies by PMUs (Phasor Measurement Units) of a WAMS (Wide Area Monitoring Systems). The distributed method of optimization and consensus used here is based on the distributed ADMM (Alternating Direction Method of Multipliers) algorithm. The performance of the distributed Prony method is evaluated by processing known analytical signals and synchrophasorial voltage measurements recorded at low voltage by the MedFasee BT Argentina Project: Observatory of the Dynamics of the Sistema Argentino de Interconexión (SADI). The results show that more precise oscillatory modes are obtained with the distributed multiProny method than with the simple Prony method when processing a single signal or a set of signals in a centralized way.