ANN based SVPWM for three-phase improved power quality converter under disturbed AC mains

This paper presents an artificial neural network (ANN)-based space vector pulse width modulation (SVPWM) control approach for better performance of three-phase Improved Power Quality Converters (IPQCs) for distorted and unbalanced AC Mains. The neural-network based controller offers the advantages of very fast implementation of the SVPWM algorithm for disturbed supply. The proposed scheme employs a three-layer feed-forward neural network which receives the command error voltage and line currents information at the input side to retransform the Clarke transformation for generating reference vector trajectory. The neural-network-based modulator retransforms the Clarke transformation to distribute the switching times for each device in each leg to have balanced line currents with nearly unity input power factor, low input current THD and reduced ripple factor of the regulated DC output voltage.