Magnitude and phase determination of harmonic currents by adaptive learning back-propagation neural network

Abstract

Harmonic currents are significant and inevitable when power electronic installations are used in industry and telecommunication energy plants. In order to compensate the instantaneous harmonic current components by active power filtering technique, it is a prerequisite to estimate the magnitude and phase of each harmonic component in real time. In this paper, a promethean approach is proposed for the determination of magnitude and phase of each current harmonic component from the distorted line currents. This approach introduces an adaptive learning multi-layer backpropagation neural network which converges faster than simple back-propagation neural network. Unlike conventional methods of harmonic current determination, this method requires only half cycle of the distorted current waves. This method is four times faster than the conventional method and this makes the on-line determination of instantaneous harmonic components.