A comparative study on the Prony analysis and the ERA for modal analysis

Abstract

The Prony analysis has been considered as a standard method for estimating oscillation modes using ringdown responses in a power grid. Extensive studies have been done to find its optimal performance conditions, but the comparisons between the Prony analysis and other modal analysis approaches from a user perspective are insufficient. This paper compares the performance of two modal analysis methods, i.e., the eigensystem realization algorithm (ERA) and the Prony analysis. Their performances are compared using a simple model and a 16-machine model. The influence of the parameters, such as the model order, the decimation factor, and signal-noise-ratio (SNR), on the modes' estimation accuracy is evaluated. Because of the randomness of noise, the Monte Carlo (MC) method is used to evaluate estimation accuracy. The median absolute deviation (MAD) is used as a metric for comparing the estimation errors. It is shown that the ERA has more preferred features than the Prony analysis in estimating power system modes.