An integrated data compression using wavelet and neural networks for power quality disturbances [articol]

Abstract

This paper introduces a novel data compression technique for the classification of power quality disturbances using wavelet transform and radial basis function neural network. For compression, criterion of maximum wavelet energy coefficient, signal decomposition and reconstruction is been used. The analysis was carried out by simulating power quality disturbance data, such as sag, swell, momentary interruptions and harmonics using MATLAB. Daubechies and Symlet functions were used to select wavelet function and scale to decompose the signals. Radial basis function neural network is been used to compress and classify various power quality disturbance for different orientations. The simulation results shows that the proposed technique compresses and classifies the signals very well when compared to conventional data compression techniques.