Power quality events detection using fourth-order spectra

This paper introduces the use of a fourth-order frequency-domain statistical estimator, the spectral kurtosis (SK), in the field of power-quality analysis. The research has been organized in the frame of a research national project and points towards the implementation of these techniques into an automatic platform to perform PQ analysis in power plants and power inverters. Higher-order statistics in the frequency domain manage to distinguish 3 types of electrical anomalies (sags, swells and transients), with an accuracy of 83%.