Analysis of current signals for diagnostics of short-circuited turns in a transformer based on Wavelet transform

Abstract

Nowadays, transformers are basic elements and of great importance in electrical systems. Unfortunately, these elements are prone to faults, among which the short-circuit fault in the windings stands out. In this regard, there is an emerging need in the proposal of methodologies that help to detect and quantify this failure in a timely manner, and avoid serious damage to the transformer or connected electrical devices in the network. The present work proposes a methodology to obtain features that help to quantify short-circuit faults by means of a Wavelet-packet transformed technique, whose function is to decompose a signal in its different frequency components. The analysis carried out is based on energizing current signals obtained experimentally in a single-phase transformer so the analysis is divided into two parts, the first part corresponds to the transient part of the signal and the second part corresponds to the analysis of the steady state of the same. The Shannon entropy is proposed as an index for the detection and quantification of the fault. Finally, various short circuit fault conditions between turns of the windings are analyzed in an experimental manner. These conditions are 10, 20, 30 and 40 short circuited turns, in addition to the healthy state of the transformer. The results obtained demonstrated the effectiveness of the proposed methodology