A novel convex combination of LMMN and RLS algorithms for power system parameter estimation

Abstract

The ever growing usage of power electronic devices and non-linear loads in a power system is increasing current and voltage harmonics which are a major source of power quality deterioration. An essential step to mitigate the negative effects of harmonics is to design proper filtering which in turn requires an efficient parameter estimation of a measured power system signal. In this paper a novel convex combination of the least mean mixed norm (LMMN) and recursive least square (RLS) algorithms called RLMMN is proposed for harmonic, sub-harmonic and inter-harmonic amplitude and phase angle estimation of a distorted power system signal corrupted with a random Gaussian noise. The proposed RLMMN algorithm inherits the advantages of both algorithms to achieve fast convergence, better tracking and lower steady state estimation errors. The estimation performance is evaluated for static signal and abrupt amplitude drift condition. Three other algorithms reported in the literature are considered for judging the robustness of the proposed algorithm. From the simulation studies it is observed that the proposed RLMMN algorithm exhibits superior estimation performance in terms of estimation accuracy as compared to estimations obtained by LMS, RLS and LMMN algorithms under various SNR environments.