Análise qualitativa das famílias Wavelet para detecção de descargas parciais em isoladores

Abstract

Partial discharges (PDs) are undesirable phenomena that can manifest themselves in the equipment that make up the electrical system, and these dielectric breakdowns tend to evolve into short circuits. In order to diagnose the damage caused by PDs in a certain equipment, the energy concessionaires need to take it out of operation and then apply the electrical method. In addition to this, the thermovision method can be used, which only detects damage to the equipment when the ruptures are already severe. This article aims to use the Wavelet Transform(TW) in the processing of information contained in 26 PDs and noise signals, Corona-type with and without saline pollution. As a result, the detection of PDs is performed based on the TW coefficients. For this, it was necessary to vary the TW parameters and find the Wavelet family and level that best suits the PD signals. The TW proved to be suitable for the treatment of ultrasound signals as it is capable of dividing a signal into time and frequency windows. This feature makes it possible to identify whether the ultrasound signal comes from noise that is normal for system operation or from a PD. With this work, it was possible to identify the most suitable Wavelet families for the identification of partial discharges in medium voltage insulators.