Adaptive neurofuzzy inference system least-mean-square-based control algorithm for DSTATCOM

Abstract

This paper proposes the real-time implementation of a three-phase distribution static compensator (DSTATCOM) using adaptive neurofuzzy inference system least-mean-square (ANFIS-LMS)-based control algorithm for compensation of current-related power quality problems. This algorithm is verified for various functions of DSTATCOM, such as harmonics compensation, power factor correction, load balancing, and voltage regulation. The ANFIS-LMS-based control algorithm is used for the extraction of fundamental active and reactive power components from nonsinusoidal load currents to estimate reference supply currents. Real-time validation of the proposed control algorithm is performed on a developed laboratory prototype of a shunt compensator. The real-time performance of shunt compensator with ANFIS-LMS-based control algorithm is found satisfactory under steady-state and dynamic load conditions. The performance of the proposed control algorithm is also compared with fixed-step LMS and variable-step LMS (VSLMS) to demonstrate its improved performance.