Shifting Window Average Method for Phasor Measurement at Offnominal Frequencies

Abstract

A shifting window average method (SWAM) is proposed for phasor measurement at offnominal frequencies. The proposed SWAM is not only applicable for fundamental frequency phasor measurement, but also for harmonic phasor measurement as well. SWAM is based on a mathematical analysis of the errors induced in the commonly used DFT method at offnominal frequency inputs. First, a comprehensive derivation of magnitude and phase-angle measurement errors of discrete Fourier transform (DFT) is proposed. It is indicated that the errors are caused by modulations between different exponential components in the signal, and can be modeled by quasisinusoids with frequencies equal to the frequency differences between modulated components. SWAM is then introduced to eliminate the errors by examining the analytic form of the errors. Simulation tests have been performed on several benchmark signals. Simulation cases have validated that SWAM achieves higher accuracy compared with the DFT-based method and the three-sample average filter method. Experimental tests were also performed to validate the accuracy of the proposed method. Due to its high accuracy and reasonably low processing effort, SWAM is a valuable candidate for online phasor measurement in power systems.