EVALUATION OF MATRIX PENCIL ALGORITHMS APPLIED IN ELECTRIC POWER SYSTEMS

Abstract

Electric power systems sometimes experience transient situations in which their steady state behavior is no longer respected. In these scenarios, the electric power supply can be compromised as well as economic losses may occur. In order to better predict such accidents, system identication uses several methods to dene signal parameters. One method is the denition of frequency-dependent network equivalent (FDNE) models to analyse eletromacnetic transitories. Another method is the denition of transient signals' eletromechanic modes based on Wide Area Monitoring (WAM) systems. An important technic of parameter identication is the Matrix Pencil method, which can be applied in both problems described. In this work, it is proposed to study and evaluate the performance of the Matrix Pencil algorithm in the WAM problem and in the estimation of eletromechanic modes. It is also purposed to compare it with the Vector Fitting method, which has been used in similar applications. Three study cases are presented in order to evaluate and compare the problem described: One synthetic test signal and two real cases considering ringdown data extracted from the North American Eastern Interconnection system and from the Brazilian Interconnected Power system.