Accurate and Fast Parallelized Assessment of Waveform Distortions in Presence of Low and High frequency Spectral Components

Abstract

The waveforms distortion assessment nowadays requires great research efforts due to the contemporaneous presence of low- and high-frequency harmonics and interharmonics (up to 150 kHz). In this paper, a new multi-method approach allowed accurate and fast analysis of such waveforms. The method profitably integrates the performances of some basic methods: the Discrete Wavelet Transform, the sliding-window modified ESPRIT method, and the sliding-window Discrete Fourier Transform which uses a Nuttal window whose length is accurately calculated thanks to the application of the Prony method or the Particle Swarm Optimization. The accurate calculation of Nuttal window length guarantees a good performance in the high-frequency spectral components assessment while reduced computational efforts are obtained parallelizing the calculation of low- and high-frequency harmonics/interharmonics. Numerical applications on synthetic and measured waveforms demonstrate that the proposed method improves the performances of high-performing methods proposed in the recent relevant literature, in the examined cases.