Fundamental frequency estimation by an interpolated DFT algorithm eliminating negative-image component interference in arbitrary windows

Abstract

Interpolated discrete Fourier transform (IpDFT) is an effective algorithm for spectrum analysis. However, the interference effect from the negative-image component significantly influences the accuracy of spectrum analysis for a small number of acquired sine wave cycles. In this study, a high-accuracy IpDFT algorithm is proposed for fundamental frequency estimation, wherein the interference effect of the fundamental negative-image component is theoretically eliminated. The proposed IpDFT algorithm uses the real or imaginary parts of three DFT bins with the largest amplitudes to establish the interpolation formula, which is employed to correct the frequency deviation caused by the picket fence effect. Additionally, this algorithm is suitable for arbitrary windows. The calculation process is simplified using the lookup-table method. Moreover, the effectiveness of the proposed IpDFT algorithm is verified by simulation and experimental results. The frequency estimation accuracy of the proposed IpDFT algorithm is comparatively analyzed with those of several state-of-the-art IpDFT algorithms for both noisy and noisy-and-harmonically distorted sine waves.