Application of Wavelet Entropy and Adaptive Nerve-fuzzy Inference to Fault Classification

A new method of fault classification based on transient signals for high voltage power transmission line is presented. Based on the wavelet analysis and the idea of entropy, the concept and definition of wavelet energy entropy (WEE) for feature extraction are introduced. First, appropriate wavelet decomposition of acquired post-fault signal is conducted and the WEE is calculated. ; then the characteristic vectors for post-fault signal are built; the characteristic vectors containing fault information are used as the input of adaptive nerve-fuzzy inference system; finally, the fault classification is realized according to the output of adaptive nerve-fuzzy inference system. The simulation results of fault classification have been analyzed on a 500 kV transmission line PSCAD/EMTDC model. The result shows that the variation regularity of WEE with time can effectively reflect fault features. With the help of adaptive nerve- fuzzy inference system, it achieves good classification effect.