Beyond FFT algorithm in analyzing harmonics at frequency range of 2 kHz to 500 kHz

Abstract

In this work, the algorithms Exact Model Order Estimation of Signal Parameters via Rotational Invariance (EMO-ESPRIT) and Sliding-Window Exact Model Order-ESPRIT (SW-EMO-ESPRIT) are analyzed in order to detect the frequency components of a distorted signal in the 2-500 kHz frequency range, with the aim of providing a possible measurement solution effective for all-over the frequency range of interest. Indeed, some standards, like IEC 61000-4-7 and IEC 61000-4-30, deal with measurement methods which are suitable for only part of that frequency range (e.g., IEC 61000-4-7: 2-9 kHz and IEC 61000-4-30: 2-150 kHz), and suggest to apply FFT algorithm on a measured rectangular window with 200 ms duration (50/60 Hz power systems). The previously mentioned methods and the method suggested by the standards are tested here on both simulated synthetic signals and experimental measured signals and the obtained results are then compared to find their 'pros and cons'.