Detection and classification of power quality disturbances using discrete wavelet transform and wavelet networks

Abstract

A novel approach for detection and classification of power quality (PQ) disturbances is proposed. The distorted waveforms (PQ events) are generated based on the IEEE 1159 standard, captured with a sampling rate of 20 kHz and de-noised using discrete wavelet transform (DWT) to obtain signals with higher signal-to-noise ratio. The DWT is also used to decompose the signal of PQ events and to extract its useful information. Proper feature vectors are selected and applied in training the wavelet network classifier. The effectiveness of the proposed method is tested using a wide spectrum of PQ disturbances including dc offset, harmonics, flicker, interrupt, sag, swell, notching, transient and combinations of these events. Comparison of test results with those generated by other existing methods shows enhanced performance with a classification accuracy of 98.18%. The main contribution of the paper is an accurate (because of proper selection of feature vectors), fast (e.g. a new de-noising approach with proposed identification criterion) and robust (at different signal-to-noise ratios) wavelet network-based algorithm (as compared to the conventional wavelet-based algorithms) for detection/classification of individual, as well as combined PQ disturbances.