Combined Two-Point and Three-Point Interpolated DFT for Frequency Estimation

In this paper, a novel methodology for frequency estimation is proposed. It is based on a root-mean-squared combination of the two-point and three-point Interpolated Discrete Fourier Transform (IpDFT) with a rectangular window. The advantages of the proposed modified IpDFT are its short window length and high accuracy in frequency estimation as compared to the conventional two-point IpDFT and three-point IpDFT methods. The effects of harmonics and noise in the measurement signal (the proposed technique is used for frequency estimation of this signal) on the proposed technique are also studied. The proposed method is tested on computer-generated test signals. The numeric results obtained prove the efficacy and accuracy of the proposed technique.

I. Introduction

Frequency estimation utilizing phasor information is essential for power system monitoring and protection. The estimated frequency may be used to activate frequency-dependent protective relays. The most widely used algorithm for this purpose is the Discrete Fourier Transform (DFT) due to its simplicity. However, DFT works well only for stationary signals. Since the voltage and current signals in the power network are usually non-stationary, DFT, generally, results in significant phasor and frequency errors for non-stationary signals.