Accurate estimation of grid voltage parameters using singular value decomposition technique

Abstract

This paper presents the comparison of results between the singular value decomposition (SVD) and the least square (LS) techniques and also proposes the application of the SVD technique for extracting the parameters of the grid voltage waveform. With a small size window, the proposed SVD technique can extract amplitudes and phase angles of all the frequency components present in the Prony modelled grid voltage waveform more accurately than the LS technique. Moreover, the proposed technique has the ability to estimate the grid voltage parameters accurately from the singular matrix, where the LS technique fails. The SVD technique is applied on the Vandermonde matrix formed by the known damping factors and frequencies of the Prony modelled grid voltage waveform in order to estimate the amplitudes and phase angles. Moreover, an algorithm, which avoids the rooting of higher order polynomial, is presented to estimate the unknown harmonic frequencies from the harmonically distorted grid voltage waveform, where the fundamental frequency is estimated from a predicted fundamental frequency variation zone. Synthetically generated grid voltage waveforms are used in MATLAB to depict the superior performance of the proposed SVD technique over the LS technique.