Bayes method of power quality disturbance classification

With the proliferation of nonlinear loads, more attention has been given to power quality problems. In order to mitigate the influence, various power quality disturbances must be classified before an appropriate measure can be taken. Wavelet packet is developed on the basis of wavelet transform, which can provide more plenteous time-frequency information. This paper uses wavelet packet to decompose power quality disturbance signals, then selects energy and entropy of terminal nodes through wavelet packet decomposition as eigenvector respectively, and uses Bayes classifier to classify the disturbances, which are simulated and analyzed. The simulation results validate the effectiveness of this method. It's showed that the entropy acted as eigenvector has higher recognition accurate ratio compared with Fisher piecewise linear classifier.