ASSESSING NONLINEARITY AND NON-STATIONARITY IN POWER SYSTEM BEHAVIOR USING HIGHER ORDER STATISTICAL ANALYSIS TECHNIQUES

Abstract –A novel analysis framework based on nonlinear statistics, for detection and quantification of nonlinear power system behavior is presented. The technique uses output-only measurements for the extraction of nonlinear characteristics of systems subjected to large perturbations and can be used to determine the strength and distribution of nonlinear behavior of both, ambient and power system response data. First, linear statistic analysis techniques are used to determine critical machines following the occurrence of severe perturbations. Then the method of surrogate data is utilized to detect nonlinearities and non-stationarities in the underlying process. Criteria for determining possible nonlinearities with respect to variables are proposed and statistical significance tests are used to give a quantitative measure of nonlinearity. Finally, higher order spectral analysis techniques are employed to investigate the nature and extent of nonlinear interaction involving the electromechanical modes of oscillation of the system. A detailed case study of a 68-bus, 16-machine test system is presented to illustrate the use of the proposed techniques. Examples are given of the application of statistical criteria to detect nonlinearity in power system signals, and estimating the strength of nonlinear interaction