Improved SDFT algorithm based on total least squares for frequency estimation in three-phase power system

In the smart discrete Fourier transform(SDFT)algorithm, the underlying relationship among the three consecutive fundamental components of the voltages does not hold when the three-phase power system is contaminated by noises, harmonics, or encountered with sudden interrupts. To solve this problem, a total least squares SDFT(TLS-SDFT)algorithm is put forward. In the proposed algorithm, the original three point relation formula in the SDFT algorithm is extended based on the multiple DFT fundamental observations obtained by sliding windows.A perturbation matrix is introduced. The coefficient matrix is singular value decomposed to minimize the Frobenious norm of the perturbation matrix, and then the estimated frequency is improved. Due to the special structure of the coefficient matrix, the additional complexity of the proposed algorithm is a linear function with the length of the sliding window. The simulation results show that the estimation bias and the mean square error of the proposed algorithm are much smaller than those of the original SDFT algorithm under the interference of Gauss white noise. The frequency tracking accuracy of the proposed algorithm is obviously improved under the conditions of high harmonic interference, signal parameter mutation and real substation measurement. © 2017, Editorial Department of Journal of Southeast University. All right reserved.