Arctangent Function-Based Normalized Least Mean Squares Algorithm for a Battery Supported Wind Energy Conversion System

Abstract

Wind energy conversion systems (WECSs) based on doubly fed induction generator (DFIG) are often connected to local loads, and excess power is supplied to the grid. The nonlinear and unbalanced local loads integrated with WECS, degrade the power quality by injecting harmonics and negative sequence components in grid currents. Another major concern in grid-integrated WECS is related to sudden steep fluctuations in wind and load powers, which bring about frequency and voltage deviations that ultimately affect grid stability. Thus, the objective of this work is twofold. First, to suppress harmonics and negative sequence components in grid currents, using a normalized least mean squares (NLMS) adaptive filtering scheme based on arctangent cost function (Arc-NLMS). The Arc-NLMS adaptive algorithm is robust against model uncertainties and exhibits optimal convergence performance. Second, to guarantee grid security amidst sudden erratic variations in wind speed or load power, through implementation of a power management scheme (PMS) using a battery energy storage. The PMS allocates exponential values to weight components, to diminish the effects of power fluctuations. Based on performance evaluation of the system using a developed laboratory prototype, the PMS proves to be effective in smoothening power fluctuations while the power quality issues related with connection of local nonlinear and unbalanced loads are also alleviated.