An Adaptive Step-Size Least Mean Square Algorithm for Electric Power Systems Frequency Estimation in protective relays

Abstract

This paper proposes a new adaptive step-size Least Mean Square Algorithm (LMS) for electric power systems frequency estimation. The algorithm is simple, computationally efficient, and makes the correction of the signal that enables to reach the mean square error. The proposed algorithm has a new kind of step adaption for LMS algorithm that provides high precision in steady state condition using small values of steps, and small time convergence using bigger values in transients. The method's performance was evaluated by indicator such as mean square error and convergence time. The tests were accomplished considering many synthesized and simulated signals with noise, harmonics, voltage and frequency variations.