Neural network controller for five phases shunt active power filter applied for five phase embarked electrical network

Abstract

This paper discusses neural network ANN controller applied for five phase shunt active power filter. Three-phase shunt active power filters are widely studied, however in this paper we investigate to extend their application to five phase embarked electrical network. Two-level five phase inverter is used as a shunt active power filter to reduce harmonics and compensate reactive power injected by a five-phase diode rectifier. An artificial neural network controller has been used to regulate the DC side of the inverter, thereby contributing to the calculation of the reference harmonic currents on the one side and controlling the injected harmonic currents on the other. Self-tuning filter (STF) is used to calculate the reference current. By using STF allow to extract the voltage and current fundamental components in the α-β-axis directly without the use of a phase-locked loop (PLL). Simulation results shows that the ANN controller presents appreciable performances in references tracking and the proposed five-phase shunt active power filter provides a sinusoidal supply current in phase with the line voltage with low harmonic distortion.