Estimation of power quality indices in distributed generation systems during power islanding conditions

Abstract

This paper presents a new, fast Modified Recursive Gauss–Newton (MRGN) method for the estimation of power quality indices in distributed generating systems during both islanding and non-islanding conditions. A forgetting factor weighted error cost function is minimized by the well known Gauss–Newton algorithm and the resulting Hessian matrix is approximated by ignoring the off-diagonal terms. This simplification produces a decoupled algorithm, for the fundamental and harmonic components and results in a large reduction of computational effort, when the power signal contains a large number of harmonics. Numerical experiments have shown that the proposed approach results in higher speed of convergence, accurate tracking of power signal parameters in the presence noise, waveform distortion, etc., which are suitable for the estimation of power quality indices. In the case of a distribution network, power islands occur when power supply from the main utility is interrupted due to faults or otherwise and the distributed generation system (DG) keeps supplying power into the network. Further, due to unbalanced load conditions the DG is subject to unbalanced voltages at its terminals and suffers from increased total harmonic distortion (THD). Thus, the power quality indices estimation, along with the power system frequency estimation will play a vital role in detecting power islands in distributed generating systems. Extensive studies, both on simulated and real, benchmark hybrid distribution networks, involving distributed generation systems reveal the effectiveness of the proposed approach to calculate the power quality indices accurately.