A New Restricted Earth Fault Relay Based on Artificial Intelligence

Abstract

The restricted earth fault (REF) relay is a type of differential protection which is used for detection of internal ground faults of power transformers. But, during external faults and transformer energization conditions, the probability of current transformer (CT) saturation increases. Thus, the spurious differential current due to CT saturation, can lead to REF relay maloperation. In this paper, a new intelligent REF protection scheme is presented based on the pattern recognition method. In the proposed method, one-cycle data window of differential and neutral currents are used as exemplar patterns of the classifier. The performance of the proposed method is evaluated by obtained data from simulation of a real 230/63 kV power transformer in PSCAD/EMTDC environment. Moreover, in order to accurately simulate the CT behavior during saturation the well-known Jiles-Atherton (JA) model is utilized. The promising obtained results showed that the proposed intelligent method increases the security of the REF relay.