Power quality disturbance waveform recognition using wavelet-based neural classifier - Part 2: Application

A wavelet-based neural classifier is constructed and thoroughly tested unde r various conditions. The classifier is able to provide a degree of belief for the identified waveform. The degree of belief gives an indication about the goodness of the decision made. It is also equipped with an acceptance threshold so that it can reject ambiguous disturbance waveforms. The classi fier is able to achieve the accuracy rate of more than 90 % by rejecting le ss than 10% of the waveforms as ambiguous.