Prevention of Maloperation of the REF Relays Using Artificial Intelligence and Wavelet Transform

Restricted Earth Fault (REF) Relay, a type of differential protection, is responsible for detecting single-phase-to-ground faults at the terminal of a transformer or faults in the winding to the core. This protection scheme is inherently sensitive and rapid; however, due to current transformer saturation during external fault conditions with high currents and the transient magnetizing current flowing through the transformer core, it may be susceptible to false operation. In this paper, an intelligent restricted ground fault protection scheme based on wavelet transform (WT) is presented. In the initial step, the differential current resulting from the simulation is analyzed using WT, and various features of decomposed signals are extracted as sample patterns for training intelligent classifiers. The performance of the proposed method is evaluated using data obtained from the simulation of a 230/63.5 kV power transformer in the PSCAD/EMTDC software environment. Additionally, to accurately simulate the transformer current during saturation, the valid Gill's model is employed. The results of implementing this intelligent protection scheme confirm its high reliability against false operation.