Comparison of Matrix Pencil and Prony methods for power system modal analysis of noisy signals

Abstract

Modal information extracted from the dynamic response of power systems can be applied to detect low frequency oscillations and assess stability margins for monitoring and preventive control. This paper examines two techniques for modal identification based on their ability to accurately identify system modes in the presence of noisy signals. The methods investigated include Prony analysis, which has commonly been used in power system studies, and the Matrix Pencil method, which is more common in electromagnetic analysis. Prony analysis has been shown to have difficulties extracting the modes of noisy signals, so the examples presented explore these shortcomings and compare them to the capabilities of the Matrix Pencil method.