Neuro-fuzzy based controller for a shunt active power filter

Abstract

This paper describes the application of a novel neuro-fuzzy based control strategy which is used in order to improve the Active Power Filter (APF) dynamics to minimize the harmonics for wide range of variations of load current under various conditions. To improve dynamic behavior of a shunt active power filter and its robustness under range of load variations, instantaneous p-q theory is used with the inclusion of neural network filter for reference current generation and fuzzy logic controller for DC voltage control. The proposed control scheme is simple and also capable of maintaining the compensated line currents balanced, irrespective of unbalancing in the source voltages. The results presented in MATLAB-SIMULINK package in this paper clearly reflect the effectiveness of the proposed APF to meet the IEEE-519 standard recommendations on harmonic levels.