PROJECT TITLE :

Implementation of Echo-State Network Based Control for Power Quality Improvement - 2017

ABSTRACT:

This paper proposes a replacement control algorithm for mitigation of power quality issues at the distribution level. The control algorithm is intended on the premise of echo-state network, which is a kind of recurrent neural network. This technique is used to extract elementary weight parts from nonsinusoidal load currents. These weight components are used to guage reference grid currents and consequently generate switching logic for the voltage source converter utilized in the active filter. The shunt-connected active filter with the proposed control algorithm is employed to beat several current-related power quality issues and provides the load compensation. Problems like the presence of lower order harmonics, insufficient reactive power, and load unbalancing are corrected using a shunt filter in both the facility factor correction and voltage regulation modes. The proposed management algorithm is validated on a true time system that could be a prototype of a shunt filter. The algorithm is tested in MATLAB using Sim Power Systems and SIMULINK and implemented on a digital signal processor.