CONTRIBUTION TO AUTOMATIC DETECTION AND DIAGNOSIS OF WIDE VAREIETY RANGE OF POWER QUALITY DISTURBANCES USING COMBINED WAVELET TRANSFORM AND NEURAL NETWORK METHODS

摘要

In this paper a new approach for detection and classification of wide variety range (15 types) of power quality violation based on IEEE 1159 standard is presented. It involves abroad range of disturbances from low frequency dc offsets to high frequency transients or low duration impulse to steady state events. For this purpose wavelet multiresolution signal analysis is used to de-noise and then decompose the signal of power quality event to extract its useful information. After this an optimal vector (with 8 elements) of computed features is selected and adopted in learning a neural network classifier. This vector witch consists statistical parameter of frequency related detailed and approximation wavelet coefficients represents a distinctive property of studied power quality events. For neural network structure multi-layer perceptron (MLP) and radial basis function (RBF) are used and compared together. The proposed classifier can significantly improve automatically diagnosis efficiency of power quality disturbances. Simulation results with low error rate confirm the capability of proposed method.