A Simulation of Active Shunt Filter with Neural Network Control to Improve Voltage Profile and Power Quality

Abstract

Nonlinear loads lead to the creation of harmonics in power grids and a reduction in power quality. Furthermore, the nonlinear loads are not constant and change randomly. Since the quality of electric power manufactured in any country is a measure of the industrial progress of that country, methods are needed to reduce the harmonics of the power grid. Passive filters are normally used to reduce current harmonics during connecting nonlinear loads to the power grid. Whereas these filters have problems such as large aging size, resonance, and constant compensation. The progress of power electronic elements and the reduction of production cost of these elements have made active filters a suitable option for reducing harmonics and compensating reactive power to improve power quality. The purpose of this paper is to simulate active shunt filters through neural network control to improve voltage profile and power quality. The simulation results showed that the reactive power received from the source reaches “0” after one cycle, which leads to a power factor of “1” and the ideal performance of the adaptive shunt active filter with load changes. On the other hand, the dynamic performance of the filter is also ideally improved. The THD percentage of the source current is always less than five during load changes, which shows the ideal performance of the adaptive shunt active filter in the steady state.