Entropy Based Feature Extraction for Power Quality Disturbances

Abstract

Power line disturbances are mainly caused due to voltage sag, voltage swell, outage and harmonics. These disturbances must be recognized and classified. This paper presents wavelet based decomposition of the disturbances using norm entropy and pattern recognition by probabilistic neural network. In this paper total of 175 power quality disturbance signals are considered for designing the neural network model.

Key Words: Power quality, wavelet packet decomposition, entropy, probabilistic neural network.