Recognising nonlinear interactions in power systems

Abstract

The availability of high speed data acquisition systems (DAS) allows voltage and current waveforms to be captured on-line as a time series. The voltage and current oscillations, of varying periodicity, arise out of nonlinear interactions in power systems. Conventional spectral analysis is based on the assumption that the time-series data is Gaussian. This assumption, as will be shown, cannot be universally applied. Higher-order spectral (HOS) analysis, on the other hand, offers a method to recognise these interactions in a power system when the waveforms are non-Gaussian. The time series data, to be analysed in this paper, in particular, is taken from a system of well known dynamical equations in the power system. We demonstrate, by using HOS, the extent of joint dependence of the spectral components.