Distributed Prony analysis for real-world PMU data

An analysis to show the effect of sampling rate on estimation accuracy and the mitigation methods to obtain accurate estimation is presented and a distributed Prony analysis algorithm using consensus and subgradient update is proposed.

Abstract

Prony analysis has been applied in power system oscillation identification for decades. For a single PMU signal with 30 Hz sampling rate, merely applying Prony analysis cannot give accurate results of oscillating modes of power systems. This paper presents an analysis to show the effect of sampling rate on estimation accuracy and the mitigation methods to obtain accurate estimation. The methods include sampling rate reduction and multiple-signal Prony analysis. For multiple-signal Prony analysis, this paper proposes a distributed Prony analysis algorithm using consensus and subgradient update. This algorithm can be applied to multiple signals from multiple locations collected at the same period of time. This algorithm is scalable and can handle a large-dimension of PMU data by solving least square estimation problems with small sizes in parallel and iteratively. Real-world PMU data are used for analysis and validation. The proposed distributed Prony analysis shows being robust against sampling rate and generates reconstructed signals with better matching degree compared to the conventional Prony analysis for multiple signals.