Estimation of power system frequency using a time-domain method

Power system frequency is an important parameter in power quality monitoring. Due to nonlinear loads and other problems, the frequency slowly varies around its nominal value of 50Hz. To estimate the instantaneous frequency value, therefore, a dynamic signal model is required. In earlier works, Taylor series is used to model the waveform. The coefficients of the Taylor series are computed using the Discrete Fourier Transform. The method requires the computation of overlapping segments of the waveform samples. However, the direct use of the Fast Fourier Transform is redundant and requires a higher computation complexity. In this paper, the use of an algorithm to compute the successive discrete Fourier transforms is proposed that substantially reduces the computational complexity of the algorithm. Further, the proposed problem formulation results in a better convergence. Detailed simulation results are presented that confirm the validity of the approach.