Adaptive Polynomial Harmonic Distortion Compensation in Current and Voltage Transformers Through Iteratively Updated QR Factorization

Abstract

Measuring current and voltage harmonics has paramount importance for improving the power quality of distribution grids. However, the achieved accuracy strongly depends on the adopted instrument transformer. The present paper proposes an adaptive technique that enables an effective compensation of both the filtering behavior and the harmonic distortion introduced by current and voltage transformers, namely the strongest nonlinear effect at low-order harmonics. The approach is based on a flexible, linear in the parameters polynomial modeling of harmonic distortion in the frequency domain. Model complexity can be different from one harmonic to the other, and it is selected through an automatic iterative process to suit the nonlinear behavior at each specific harmonic order, while avoiding overfitting. In particular, the number of parameters is increased by progressively updating the QR factorization of the regressor matrix trough Householder reflections, until a convergence condition is reached. Experimental tests performed on an inductive VT and CT highlight the effectiveness of the approach.