Improved windowed spectral-peaks polynomial fitting algorithm and its application in power harmonic analysis

Windowed FFT algorithm is widely used for power harmonic analysis recently. However, there are some contradictions between the accuracy and computation burden in discrete spectral corrections. In this paper, the power harmonic analysis algorithm based on the least square method (LSM) and triangular self-convolution windows (TSCW) are proposed. Firstly, the signal is weighted by TSCW for reducing spectral leakage. Then, the polynomials for discrete spectral corrections, which are also adjustable by the computation accuracy, are founded based on the LSM. Finally, the computation formulas of amplitude, phase and frequency of harmonics are given based on the spectral correction polynomials. The power harmonic analysis experiments are taken under non-synchronized sampling and non-integral period truncation, including white noise and fundamental frequency changing, which verify the usefulness of the proposed method.