A comparative analysis of artificial neural network and synchronous detection controller to improve power quality in single phase system

Abstract

This paper explores the effectiveness of neural network control technique to extract the source current references from the distorted waveform of load current. Artificial neural network (ANN) control technique working under balanced single phase voltage source conditions is implemented on the shunt active power filter (APF) along with unipolar pulse width modulation (U-PWM), proportional integral (PI) controller is used for generation of gate signal and regulation of direct current (DC) voltage of the voltage source inverter (VSI). The U-PWM technique is based on simultaneously comparing a triangular high frequency carrier signal with a slow-varying regulation signal. A comparative analysis is made between the results obtained in this technique and synchronous detection technique; it is found that developed neural network controller is very much effective to enhance the power quality in a single phase system by minimising the total harmonic distortion to a level below IEEE-519. To validate the effectiveness of neural network controller, experimental results are provided using D-SPACE1104 which gives very encouraging results.