Power Quality Event Classification Using Digital Image Processing based ANN Techniques

H.-T. Yang, S.-C. Chen, and Y.-C. Lin (Taiwan)


digital image processing, power quality, artificial neural network, discrete wavelet transform, digital filtering,image-edge detecting


This paper proposes a digital image processing based artificial neural network (ANN) approach to classifying the disturbance signals of power quality (PQ) problem. The discrete wavelet transform (DWT) technique is used to obtain various scales of wavelet coefficients versus time. The time functions of the various scales of wavelet coefficients constitute a 2-dimension (2-D) time frequency diagram. Different disturbance signals of the PQ problems have distinct features in different scales of the wavelet coefficients and thus in the 2-D diagram. To make the features more prominent out of the noises riding on the signals, the digital filters and image-edge detecting schemes are used to effectively enhance the signal-to noise ratios of the 2-D diagram. The disturbance signals of the PQ events simulated include the voltage sag, swell, interruption, notch and transients. With different levels of noises added, the data derived are used as the training and testing samples of the ANN. Satisfactory results of classifying the disturbance signals are presented and discussed in this paper.

Important Links:

Go Back