Near-Perfect Time–Frequency Analysis of Power Quality Disturbances

Abstract

Time-frequency analysis is a key requirement for the detection of power quality disturbances (PQDs). Due to non-stationary nature of the disturbances and resolution restriction posed by the uncertainty principle, most signal processing techniques fail to single-handedly detect different types of disturbances with desired precision. As a solution to this problem, a near-perfect time-frequency analysis technique is presented for PQDs. It utilizes the concept of instantaneous frequency (IF)-based time–frequency representations. A carefully chosen Gaussian window in the windowed Fourier transform followed by the slight post-processing gives a high spectral concentration on the IF trajectory by eliminating unwanted frequencies. Several comparative case studies are shown, which support the effective detection and classification capabilities offered by the proposed technique.