MONITORING OF POWER QUALITY DISTURBANCES IN THE EGYPTIAN POWER NETWORK USING WAVELET BASED NEURAL CLASSIFIER

摘要

Modern power system loads include power electronic equipment, microcontrollers, non linear elements and large industrial motors. These types of load introduce a lot of Power Quality (PQ) problems into the power system. Fourier Transform (FT) has been used for long time to analyze the waveform distortion. New technique for waveform distortion analysis -Wavelet Transform (WT)- has been recently introduced. This paper holds a comparison between the two techniques for the analysis of actual four power quality disturbances that were captured from the Egyptian Network by means of Digital Fault Recorders (DFR's). These disturbances are due to: arc furnace operation, voltage sag during faults, voltage distortion because of transformer inrush current, and capacitor bank switching. The records of these disturbances were analyzed using both WT and FT techniques. A comparison is held showing the superiority of the WT technique over the FT one. Using a set of multiple detailed levels for WT as inputs to the Artificial Neural Networks (ANN's), an automatic decision making logic for the type of disturbance is determined. Encouraging results were obtained and opened a promising way to accomplish a complete on-line power quality monitoring system.