Particle swarm optimization trained artificial neural network to control shunt active power filter based on multilevel flying capacitor inverter

Abstract

Shunt Active Power Filters (SAPF) are an emerging power electronics-based technology to mitigate harmonic and improve power quality in distribution grids. The SAPF proposed in this paper is based on three-phase Flying Capacitor Inverter (FCI) with a three-cell per phase topology, which has the advantage to provide voltage stress distribution on the switches. However, controlling the voltage of floating capacitors is a challenging problem for this type of topology. In this paper, a controller based artificial neural networks optimized with particle swarm optimization (ANN-PSO) is proposed to regulate the filter currents to follow the references extracted by the method of synchronous reference frame (SRF). The simulation results showed an enhancement of the power quality with a significant reduction in the THD levels of the current source under various loading conditions, which confirms the effectiveness, and robustness of the proposed control scheme and SAPF topology.