Three-phase power system frequency estimation based on bias-compensated aware adaptive filtering algorithms

The least mean square (LMS) or normalized LMS for the adaptive filters have been extensively utilized to estimate the frequency of the three-phase power system adaptively due to their simple implementation and good convergence properties. However, these algorithms produce biased estimate in presence of noisy voltage readings as input measurement noises. This work is aim to design novel estimate methods based on the bias-compensated algorithms to resolve the problem of the input signal corrupted by noise. First, a frequency estimation method using bias-compensated NLMS algorithm is proposed to solve the input and output measurement noises with Gaussian characteristic. Second, to improve the robustness of the estimate method under non Gaussian output noise, we use the bias-compensated NLMF to estimate frequency. The estimate results of the proposed methods are compared with recently introduced LMS and LMF under different conditions. The estimate results appear that they substantially more accurate than the existing adaptive frequency estimate methods for three-phase power system under noisy input case.