Sample Value Adjustment Improves Phasor Estimation at Off-Nominal Frequencies

Abstract

This paper presents an algorithm for better phasor estimation at off-nominal frequencies. The sampling rate remains unchanged, but instead the amplitude of the sample values within a buffer is altered in order to give a signal of the same amplitude but different frequency. The re-calculation of the samples is done by second order spline interpolation. The effect is that the signal frequency appears to be changed to nominal. The algorithm requires knowledge of the frequency, which has to be determined by other means. It can only be used for filters with a finite window length, such as finite-impulse response filters. The proposed algorithm gives a maximum magnitude error of less than 0.1% for frequency deviations up to 10% from nominal. The proposed algorithm does not need any over-sampling and re-sampling to a lower sampling rate. It helps preserve the desired frequency characteristics of filters at off-nominal frequencies. This paper presents a simulation example, which uses an off-nominal test signal of 55 Hz with a 10% second harmonic. The result indicates that this test signal might cause problems for some phasor algorithms, but the proposed sample value adjustment algorithm handles this test case efficiently and complies with Level 1 of IEEE Standard C37.118-2005.