Extraction of feature based on wavelet packet transform

Abstract

As we already know that selection of proper mother wavelet plays key role for analysis purpose. Mostly used function for detection and to locate the power quality events is Daubiechie's (Db) wavelets. For extraction of feature entropy based method is used for various levels of resolution. By extracting features we could easily detect, locate and classify the eruptions caused in power quality. The pertinent of power quality issues has been lately elevated because of the extension utilization of power semiconductor impedimenta, that upshots in a voltage and current waveforms deviation. The Power Quality (PQ) scanned is encrusted by two main objectives: The advancement of PQ indicates to liberate the power supply condition and the electrical disturbances revelation such as harmonics, transients, sags, swells, notch etc., which allows willful the conditions of the power systems. Here in this paper feature vectors are extracted by using wavelet packet transform. The learning and discernment of these unusual conditions is required to ensure that the reliability and stability of the power system do not bear depletion as a result of any vital event.